Skip to content

Understanding Treaties vs. Executive Agreements in U.S. International Law

GovFacts
74 references across 22 domains

Last updated 2 weeks ago. Our resources are updated regularly but please keep in mind that links, programs, policies, and contact information do change.

On April 22, 2016, in the United Nations General Assembly Hall in New York, Secretary of State John Kerry signed the Paris Agreement for the United States while holding his granddaughter, Isabelle Dobbs-Higginson, on his lap. The picture ran everywhere. What it left out was the step that never happened.

The 1992 climate framework convention underneath the Paris deal had received the Senate’s advice and consent, but for Paris itself the Obama administration took the position that it was an executive agreement for which senatorial or congressional approval was not required. President Obama signed an instrument of acceptance on August 29, 2016, which was deposited with the UN Secretary-General days later. In January 2025 a successor ordered the U.S. Ambassador to the United Nations to submit formal written notification of the United States’ withdrawal.

So when the news says the United States signed an agreement, did the Senate vote on it? Almost certainly not. More than 90 percent of America’s international agreements are executive agreements rather than treaties, and the label the executive branch picks decides everything that follows: whether a judge will enforce it, whether it overrides your state’s law, and how fast the next president can undo it.

The two-thirds rule, and the queue behind it

The constitutional text is short. Article II gives the President power to make treaties by and with the advice and consent of the Senate, and only if two thirds of the Senators present concur. Executive agreements appear nowhere in it. The whole apparatus of executive agreements grew up around a clause that does not acknowledge it exists.

To see what a two-thirds threshold does in practice, look at what it has stopped. The State Department keeps a running list of treaties that have been sent to the Senate and have never received advice and consent to ratification. The oldest entry, as of the list’s January 2025 update, is International Labor Organization Convention No. 87 on freedom of association, adopted at San Francisco in 1948 and submitted to the Senate on August 27, 1949. Nobody withdrew it. Nobody voted it down. It simply sat.

Treaties with powerful champions die too. The Convention on the Rights of Persons with Disabilities had the backing of President Obama and of former Senate Majority Leader Bob Dole, and on December 4, 2012, the Senate rejected it by 61 yeas to 38 nays, five votes short of the two-thirds majority needed, as dozens of Senate Republicans objected that it would create new abortion rights and impede the ability of people to homeschool disabled children. A president who watches that happen learns something about which door to use next time.

Four labels, and what each one buys

There is one treaty form and three kinds of executive agreement, sorted by where the President’s authority comes from. A congressional-executive agreement is one Congress authorizes through legislation enacted through the bicameral process, meaning both chambers vote. An executive agreement made pursuant to a treaty rests on a treaty the Senate already approved. A sole executive agreement rests on nothing but the President’s own constitutional powers.

Then there is a fifth thing, which is not an agreement at all. When Representative Mike Pompeo asked what the Iran nuclear deal legally was, Assistant Secretary for Legislative Affairs Julia Frifield replied on November 19, 2015, that it is not a treaty or an executive agreement, is not a signed document, and reflects political commitments. A political commitment creates no legal obligation that anyone could breach.

How much of the country’s business runs through each channel depends on who is counting and over what years, and the differences are the interesting part.

Who countedWhat was countedFigure
Commentators, as reported by the Congressional Research Service (CRS)Share of all U.S. international agreements taking executive-agreement formmore than 90 percent
A study cited by Oona Hathaway, covering 1946 to 1972Executive agreements resting at least partly on statutory authority88.3 percent
Same studyExecutive agreements resting on a treaty6.2 percent
Same studySole executive agreements, resting on the President’s own constitutional powers5.5 percent
Curtis Bradley and Jack GoldsmithSole, treaty-based and ex ante congressional-executive agreements combinedclose to 94 percent of all binding U.S. agreements

The middle rows cut against the usual story. In that period 88.3 percent of executive agreements were based at least in part on statutory authority, and only 5.5 percent were sole executive agreements. The President acting entirely alone is the rare case. What Congress mostly did was hand the authority over in advance, by statute, and then stop watching.

Curtis Bradley and Jack Goldsmith put the three presidentially controlled forms together at close to 94 percent of all binding U.S. agreements, and quote Hathaway’s conclusion that the work of agreement-making has come to be borne almost entirely by the President alone.

The 1955 circular that picks the label

Somebody has to decide which form a given commitment takes, and that somebody works at the State Department. The original Circular 175 was a 1955 Department Circular prescribing prior coordination and approval of treaties and international agreements, written to keep treaty-making within constitutional and other appropriate limits. It is not a statute. It is the department’s own housekeeping rule, and it has been answering this question since 1955.

The Foreign Affairs Manual now lists the factors officials weigh. Among them: the extent to which the agreement involves commitments or risks affecting the nation as a whole, whether it is intended to affect state laws, whether it can be given effect without later legislation by Congress, past practice on similar agreements, the preference of Congress, the degree of formality desired, the proposed duration and the need for prompt conclusion, and general international practice.

Read that list again and notice what it is. These are prudential considerations, not a line. Nothing in them tells an official that a commitment above a certain size must go to the Senate, and the factor about the preference of Congress sits alongside the factor about the need to move quickly, with no rule about which wins.

The paperwork is ordinary. Authority to negotiate or conclude arrives as an action memorandum cleared with the Office of the Legal Adviser, including the Assistant Legal Adviser for Treaty Affairs, plus the Assistant Secretary for Legislative Affairs and any agency with primary responsibility or a substantial interest.

Full powers to sign a treaty headed for the Senate cannot be delegated, because international law lets only heads of state and foreign ministers issue them. But by the department’s own account the more typical Circular 175 request goes no higher than an official at the Assistant Secretary level or above.

If you are wondering whether a court will second-guess that choice, it will not. When the Made in the USA Foundation argued that the North American Free Trade Agreement (NAFTA) was void because it had never been submitted to the Senate as a treaty, a federal court of appeals concluded that the question of what form an international agreement should take was a nonjusticiable political question. The label is chosen inside the executive branch, under a procedure the executive branch wrote for itself, and reviewed by nobody outside it.

A treaty the Senate approved that saved no one

Suppose the Senate does give its advice and consent. Does that hand you something you can use in court? Often it does not.

José Ernesto Medellín was one of 51 named Mexican nationals whom the International Court of Justice found had not been informed of their rights under Article 36(1)(b) of the Vienna Convention on Consular Relations, a treaty the United States had ratified. The International Court of Justice held that those individuals were entitled to review and reconsideration of their U.S. state-court convictions and sentences regardless of state rules that would normally bar a late challenge. Medellín asked the Texas courts to reopen his case on that basis.

The U.S. Supreme Court said the treaty did not do that work by itself. A treaty may constitute an international commitment, the Court wrote, but it is not binding domestic law unless Congress has enacted statutes implementing it, or the treaty itself conveys an intention that it be self-executing and is ratified on that basis. This is the single most useful thing to understand about Senate approval. It makes a promise to other countries. It does not automatically make a rule that a judge in Texas has to follow.

The President had tried to close the gap with a memorandum directing state courts to comply. The Supreme Court held that this did not work either, because a President’s authority to act must stem either from an act of Congress or from the Constitution itself. The United States argued that Congress had acquiesced over the years; the Court answered that acquiescence is pertinent when the President’s action falls within the second Youngstown category, not the third, and that in any event none existed here.

On the night of August 5, 2008, after a divided Supreme Court refused a reprieve, Medellín was executed in Texas. He apologized to those gathered to watch him die and told them never to harbor hate. Nine minutes later, at 9:57 p.m., he was pronounced dead. A ratified treaty, a judgment of an international court in his favor, and a presidential memorandum ordering compliance were all on the table, and not one of them was domestic law.

The agreement nobody voted on that erased a lawsuit

Now run it the other way. Can a commitment nobody voted on reach into a case you have already filed?

Dames and Moore’s wholly owned subsidiary held a written contract with the Atomic Energy Organization of Iran to conduct site studies for a proposed nuclear power plant, and Iran terminated that contract for its own convenience on June 30, 1979. The company said it was owed $3,436,694.30 plus interest for work already done. It sued and obtained orders of attachment against Iranian bank property, the legal step that freezes assets so a winning plaintiff has something to collect from.

Then the hostage crisis ended. Under the Algiers Accords the United States undertook to terminate all legal proceedings in United States courts involving claims of U.S. persons against Iran, to nullify all attachments and judgments obtained in them, and to send the claims to binding arbitration instead. The attachment the company had won was wiped out by executive order. No vote of any kind preceded it.

The Supreme Court upheld the President, and its reasoning rested less on any single statute than on history. The United States, the Court said, has repeatedly exercised its sovereign authority to settle the claims of its nationals against foreign countries, sometimes by treaty but also through a longstanding practice of executive agreements made without the advice and consent of the Senate. Practice, repeated long enough and left unprotested by Congress, had become authority.

The same logic runs back further. A Russian corporation, Petrograd Metal Works, had deposited money before 1918 with the New York private banker August Belmont. In 1918 the Soviet government dissolved, terminated, and liquidated the corporation and nationalized all its property wherever situated, including that account.

When the United States later claimed the deposit under an exchange of letters with the Soviet government, the Supreme Court held that the assignment and the agreements made with it did not, as in the case of treaties, require the advice and consent of the Senate.

It reaches state legislatures too. California’s Holocaust Victim Insurance Relief Act of 1999 required any insurer doing business in that State to disclose information about all policies sold in Europe between 1920 and 1945 by the company itself or anyone related to it.

The Supreme Court held that the statute interfered with the national government’s conduct of foreign relations, principally an executive agreement with Germany establishing a foundation to handle such claims, and was therefore preempted. A law passed by an elected state legislature lost to a document the Senate never saw.

There is a floor, and it is the Constitution itself. In Reid v. Covert the Supreme Court said the obvious and decisive answer is that no agreement with a foreign nation can confer power on the Congress, or on any other branch of government, which is free from the restraints of the Constitution. An agreement can beat your state’s law. It cannot beat your constitutional rights.

Can the next president just walk away?

Mostly yes, and the question that actually binds is usually the agreement’s own exit clause rather than anything in the Constitution.

Paris is the clean example. Under Article 28.2, a notice of withdrawal does not become effective until one year after the UN Secretary-General receives it, and because the agreement did not enter into force until November 4, 2016, the United States could not fully withdraw under that procedure until November 4, 2020.

The January 2025 executive order, by contrast, said the United States would consider its withdrawal and any attendant obligations to be effective immediately on giving notice. Those two statements are hard to reconcile, and the space between them is the whole difference between what a treaty text says and what a president decides to do.

Senate-approved treaties are not obviously safer. The 1954 Mutual Defense Treaty with the Republic of China said it would remain in force indefinitely and that either party could terminate it one year after giving notice. President Carter gave that notice without asking the Senate, Senator Barry Goldwater and twenty-four other members of Congress sued, and a federal court of appeals ruled in favor of the president.

The Supreme Court then declined to settle it. It granted the writ, vacated the court of appeals judgment and remanded the case to the District Court with directions to dismiss the complaint, four Justices treating the dispute as a nonjusticiable political question and Justice Powell finding it unripe. So the question of whether a president may end a ratified treaty alone has never been answered by the Court. It has simply been done, and not successfully challenged.

The executive branch’s internal position is blunter. Advising on the Anti-Ballistic Missile Treaty on November 15, 2001, Justice Department lawyers John Yoo and Robert Delahunty wrote that the President’s power to suspend treaties is wholly discretionary and may be exercised whenever he determines that it is in the national interest to do so.

Congress is not powerless, but its lever is domestic. It can repeal or amend the implementing legislation, and CRS notes that such a repeal could encourage the President to withdraw from the pact, while in other cases the implementing statute itself dictates what happens to domestic law once the underlying agreement ends. That is real power over the American half of the ledger and none at all over the international half.

What Congress is supposed to be told

Congress’s answer to being cut out was disclosure rather than approval. Legislation first enacted in 1972, commonly called the Case-Zablocki Act and codified at 1 U.S.C. 112b, requires the Secretary of State to transmit to Congress the text of all executive agreements within 60 days after the agreement enters into force. In plain terms: you cannot stop it, but you have to be told.

The rule was tightened in the National Defense Authorization Act for fiscal 2023. Rather than reporting only after entry into force, which can come long after an agreement is negotiated and signed, the department must now report both when an instrument is signed, concluded, or otherwise finalized and again when it enters into force.

The same section requires the Secretary to publish the text, authorizing authority, and implementing authority for each instrument on the department’s website within 120 days, and it removed the Secretary’s old power to decide which instruments held insufficient public interest to publish.

Then came the audit. The Government Accountability Office (GAO) found that State reported 311 agreements and instruments to Congress between October 2023 and March 2025, and that nearly one-third were reported late at one or both of the two required points, running on average 2.3 reporting cycles behind. GAO’s worked example: an agreement concluded on November 4, 2024 was due by the end of December 2024, was not reported until the end of March 2025, and so reached Congress about five months after it was made.

Publication fared no better. GAO found that the department missed the 120-day statutorily mandated deadline about half of the time, and that it does not track whether its own bureaus and other agencies submit on time, so it cannot say where the delays begin.

Nothing much happens when a deadline slips, which is part of the explanation. The statute’s notes record a funding restriction that restricted the use of funds during fiscal years 2005, 2006, and 2007 to implement international agreements whose text had not been transmitted within 60 days.

Those fiscal years are long gone. Of GAO’s recommendations, the department concurred with five recommendations and did not concur with two.

Some agreements are not on any list at all. Writing in Lawfare, a researcher used the Freedom of Information Act to ask the State Department for evidence of secret agreements and uncovered new, previously unreported data about 61 agreements entered into between 1994 and February 2018, with their foreign parties and general subject matter. That account notes the figure is almost certainly a portion of the total, because the department would not describe agreements whose very existence is classified.

The deals nobody meant to bind

Why sign something that binds no one? The United States gave its own answer in a 2025 submission to the International Law Commission: non-binding instruments provide the flexibility to memorialize shared understandings, make political commitments, or advance cooperation with minimal procedural or other requirements, in circumstances where establishing legal rights and obligations is not necessary. Two governments can settle what they intend to do without either of them promising a court anything.

The price of that flexibility is durability, and in March 2015 a group of Republican senators put the point to Iran’s leaders in writing. The open letter, organized by freshman Senator Tom Cotton of Arkansas, warned that any agreement not approved by Congress would be nothing more than an executive agreement between President Obama and Ayatollah Khamenei, that the next president could revoke it with the stroke of a pen, and that future Congresses could modify its terms at any time.

Congress did carve out a role, though a strange one. The Iran Nuclear Agreement Review Act gave Congress a defined window after the agreement was transmitted, during which the President could not grant statutory sanctions relief; relief could then proceed if, following the period for review, no joint resolution had been enacted. Look at the default. Doing nothing let the deal proceed, which is the exact reverse of the treaty rule, where doing nothing kills it.

How NAFTA passed with sixty-one votes

There is a third door, and trade goes through it. On November 20, 1993, the Senate passed the North American Free Trade Agreement Implementation Act by YEAs 61 NAYs 38, one senator not voting, under a rule requiring only a simple majority. Nineteen years later the Senate rejected the disabilities convention on exactly that count: 61 to 38, one not voting. The same votes, the same chamber, one agreement enacted and one dead, and nothing between them but the door each came through.

The constitutional cover was written down in advance. In a November 22, 1994, opinion, the Justice Department’s Office of Legal Counsel concluded that the Uruguay Round Agreements did not require ratification by the Senate as a treaty and could constitutionally be executed by the President and approved and implemented by ordinary Act of Congress. CRS marks the 1990s as the moment the old academic argument became routine practice, when the United States joined NAFTA and the World Trade Organization through congressional-executive agreements.

Trade Promotion Authority is the machinery that makes it run. The President must notify and consult Congress before signing, then send up the agreement with an implementing bill limited to a provision approving a trade agreement and the statement of administrative action, plus only such provisions as are strictly necessary or appropriate to implement it.

That bill then gets a vote it cannot escape and cannot survive intact: no committee may sit on it past 45 days before it is automatically discharged and voted on, and no amendment to it is in order in either chamber. Congress gives up the power to change a word in return for the power to say yes or no, and the other country gets a deal that will not be picked apart on the Senate floor.

The Senate’s case: a power given away

The Senate has been making the same complaint for decades, and its own institutional history states it plainly. The Senate was concerned, that history records, that the executive branch may use executive agreements as a substitute for treaties to avoid submitting them to the Senate for advice and consent.

The complaint has flashpoints. In December 1971 the Nixon administration concluded executive agreements with Portugal and Bahrain providing for continued stationing of U.S. military personnel at a base in the Azores and continued use of support facilities in Bahrain. Several members of the Senate Foreign Relations Committee introduced S.

Res. 214, saying that any agreement with Portugal should be submitted as a treaty to the Senate for advice and consent, and in January 1972 Senator Clifford Case moved to extend the same demand to the Bahrain agreement. Military facilities in two countries, and the senators learned of it afterward.

Earlier, they tried to amend the Constitution. Senator John Bricker’s proposal, aimed at making it harder for presidents to make international agreements, and one of the two fears behind it was the growing use of executive agreements, came within one vote of succeeding in the Senate in February 1954.

The strongest modern version of the argument is about design, not pique. The Heritage Guide to the Constitution contends that claims of complete interchangeability, meaning claims that anything doable by treaty is doable by congressional-executive agreement, seem counter to the Framers’ intent: the Framers carefully considered the supermajority rule and adopted it in response to specific threats to the Union, so finding a complete alternative to the Treaty Clause would in effect eliminate that rule and make important international agreements easier to adopt than the Framers wished.

Laurence Tribe supplies the textual version, arguing that the Treaty Clause is the exclusive means for Congress to approve significant international agreements. On this view the two-thirds requirement is not an obstacle in front of the policy. It is the policy.

Proposals keep coming. The Preventing Actions Undermining Security without Endorsement Act, introduced by Senators Markey and Van Hollen on July 23, 2020, would have prohibited actions to terminate or withdraw the United States from certain international agreements without a joint resolution of approval enacted into law. That would flip the default: Congress would have to say yes before an exit took effect, rather than watch one happen.

The executive’s answer: a country has to be able to act

The executive branch does not treat any of this as a workaround, and its claim is textual too. It rests on the executive power, the commander in chief power, the treaty power and the power to receive ambassadors, all in Article II, and the Supreme Court has upheld sole executive agreements resting on those sources in Belmont, Pink, Dames and Moore, and Garamendi. When four decisions across seven decades come out the same way, an administration reads that as settled law rather than as a run of lucky escapes.

Nor, on this account, is the form chosen for convenience. The Office of the Legal Adviser’s public explanation is procedural and unglamorous: there are two procedures under domestic law through which the United States becomes a party to an international agreement.

The practical case is easiest to see in the Dames and Moore facts. Hostages were coming home in exchange for moving American claims out of American courts and into arbitration, and a government that must first secure a two-thirds Senate vote cannot make that trade on any timetable a hostage crisis allows. The same holds for the thousands of routine instruments on air services, taxation, law enforcement cooperation and defense logistics that no senator would want to spend a week of floor time on.

The scholars who would retire the Treaty Clause

A third camp says the supermajority itself is the defect. Oona Hathaway argues that the Article II treaty has been steadily losing ground and that it is time to complete the transition, replacing most remaining Article II treaties with ex post congressional-executive agreements approved by majorities of both chambers. Her argument is not really about speed. It is about failure: if a treaty is rejected, there will be no international agreement, and that gap does significant harm to the nation’s ability to engage in international cooperation.

Bruce Ackerman and David Golove press the point further, arguing the change has already happened as a matter of constitutional law. Major accords including NAFTA and the World Trade Organization are approved by simple majorities of both Houses, a modern development departing radically from the constitutional practice of the Republic’s first 150 years, which they treat as a legitimate transformation produced by a sea change in public opinion during and after the Second World War.

Notice that this camp and the Heritage Guide agree about the facts and split on the verdict. Both say the Treaty Clause has been displaced. One calls that a loss; the other calls it an amendment by other means.

The chemical weapons treaty and a neighbor’s mailbox

The last objection is the one most likely to touch an ordinary life. Carol Anne Bond obtained an arsenic-based compound and bought potassium dichromate online, and between November 2006 and June 2007 Bond went to the victim’s home on at least 24 occasions and spread the chemicals on her car door, mailbox, and door knob. The harm the Supreme Court’s opinion describes is a minor burn on the victim’s thumb. Federal prosecutors charged Bond under the statute implementing the Chemical Weapons Convention.

That was possible because of a sentence written in 1920. In Missouri v. Holland, Justice Holmes wrote that if the treaty is valid there can be no dispute about the validity of the statute under Article I, Section 8, as a necessary and proper means to execute the powers of the government. Read broadly, that means a treaty can enlarge the subjects Congress may legislate about, which is how a dispute between two neighbors ended up under a chemical weapons law.

The Cato Institute, joined by the Center for Constitutional Jurisprudence, the Atlantic Legal Foundation and Edwin Meese, filed in support of Bond, arguing that Congress’s limited and enumerated powers cannot be increased by treaties and that the contrary reading rests on a single sentence written by Holmes. Nicholas Quinn Rosenkranz, who authored that briefing, locates the limit in the Necessary and Proper Clause itself: legislation that has nothing to do with a treaty’s subject matter would be neither necessary nor proper for carrying that treaty into execution.

The Supreme Court avoided the constitutional question and freed Bond on the statute instead. It concluded that in this curious case it could insist on a clear indication that Congress meant to reach purely local crimes, found none, and so never decided whether the treaty power lets Congress legislate past its enumerated powers.

Justices Scalia, Thomas and Alito concurred only in the judgment and would have answered that question. It remains open, which means the outer boundary of what a treaty can authorize Congress to do inside the United States is genuinely unsettled.

What fixing it would cost

The reform with the broadest support is not about approval at all. Writing in the Harvard Law Review, the authors of a study of the disclosure regime propose to do away with the bifurcated regime and impose a comprehensive publication requirement, with minor exceptions, modelled on the one long in place for federal regulations, and to reorganize the executive branch’s internal collection and publishing process along the same lines. Nobody loses a power under that proposal. The country simply gets to read what it has promised.

The auditors’ version is narrower and more concrete. GAO made seven recommendations, including that State improve the timeliness and completeness of its reporting and optimize its website for discoverability and searchability, along with tracking which bureaus and agencies file late and collecting the legal authority claimed for each instrument. Those are filing-cabinet reforms, and two of them were still refused.

Which leaves you with a rule of thumb worth carrying. When you read that the United States has signed something, ask what form it took, because the form predicts everything else. A ratified treaty may still be unenforceable in an American court unless Congress wrote it into law or the text says it is self-executing.

An agreement the Senate never saw can still wipe out an attachment you obtained and preempt your state’s statute. A political commitment binds nobody and lasts as long as the signatories want it to. And each of those labels is chosen by lawyers inside one building, under a procedure that building wrote for itself.

Articles are now written and checked by the GovFacts Engine, an AI system. No government agency has any input into what it produces. Learn more about our article development and editing process.

We appreciate feedback from readers like you. If you want to suggest new topics or if you spot something that needs fixing, please contact us.