A treaty in international law is a written agreement between states, governed by international law, intended to create legally binding rights and obligations. The definition comes from Article 2(1)(a) of the Vienna Convention on the Law of Treaties, which specifies that a treaty is an international agreement concluded between states in written form and governed by international law, whether embodied in a single instrument or in two or more related instruments and whatever its particular designation. This definition is narrower than common usage suggests. A treaty is not defined by its title, its subject matter, or whether it is registered with the United Nations. It is defined by three elements: written form, state parties, and an intent to be bound under international law.
The Vienna Convention on the Law of Treaties was adopted on May 23, 1969, and entered into force on January 27, 1980. It is often called the “treaty on treaties” because it codifies the rules governing how treaties are made, interpreted, and terminated. But the Convention does not create treaties; it regulates them. A treaty can exist without the Convention, and many do. What the Convention provides is a shared vocabulary and a set of default rules that states can rely on when their agreements are silent or ambiguous. Understanding what a treaty is requires understanding what the Convention does and does not do, and why the boundary between binding and non-binding instruments is more contested than the text suggests.
The Vienna Convention Definition
Three Essential Elements
The Vienna Convention definition distills treaty status into three elements. First, the instrument must be in written form. Oral agreements between states can be binding under customary international law, but they fall outside the Convention’s scope. Second, it must be concluded between states. Agreements involving international organizations are covered by a separate 1986 convention that has not entered into force, though the principles are similar. Third, it must be governed by international law. This is the element that distinguishes a treaty from a contract governed by domestic law, such as a commercial agreement between a state and a private company.
The fourth element, intent to create legal obligations, is implicit in the definition but decisive in practice. The International Court of Justice has repeatedly held that the label attached to an instrument is not determinative. In the 2017 Maritime Delimitation in the Indian Ocean case between Somalia and Kenya, the ICJ examined a Memorandum of Understanding and concluded that it was a binding treaty because its text stated that it “shall enter into force upon its signature” and its provisions recorded binding commitments . The lesson is that a document called an MOU can be a treaty, and a document called a treaty can fail to be one if the parties did not intend to create legal obligations.
What the Definition Excludes
The Vienna Convention definition excludes several categories of instruments that are sometimes confused with treaties. It does not cover agreements between states and international organizations, which are addressed by the 1986 Vienna Convention. It does not cover oral agreements, though customary international law may still bind states that make them. Most importantly, it does not cover non-binding instruments. A joint communiqué, a declaration of intent, or a political memorandum may record an understanding between states, but if the parties did not intend to create legal rights and duties, it is not a treaty. The distinction is not always easy to draw, and the ICJ has decided several cases turning on precisely this question.
The definition also does not require a particular form or title. A treaty can be called a convention, a covenant, a protocol, a charter, an accord, or an exchange of letters. The name is irrelevant. What matters is the content and the intent. This is why the Vienna Convention definition refers to “whatever its particular designation.” A document titled “Memorandum of Understanding” may be a treaty if it uses binding language such as “shall” and “agree,” while a document titled “Treaty” may be non-binding if it uses hortatory language such as “should” and “endeavor.” Practitioners learn to read the verbs, not the title.
The Generic and Specific Senses
The word “treaty” has both a generic and a specific meaning in international law. In the generic sense, it embraces all instruments binding at international law concluded between international entities, regardless of their formal designation. The Vienna Convention uses the term in this generic sense. In the specific sense, “treaty” refers to a particular instrument that meets the Convention’s definition and is concluded with the requisite formality, often requiring ratification. The distinction matters for domestic constitutional purposes: in the United States, for example, a “treaty” requires Senate advice and consent, while an executive agreement does not, even though both are treaties in the generic international law sense.
This dual usage creates persistent confusion in public debate. When a news report says that a president signed a “treaty,” it may mean an Article II treaty requiring Senate approval, or it may mean an executive agreement that is a treaty under international law but not under U.S. constitutional law. The Vienna Convention’s definition is the international law baseline, but domestic legal systems layer their own categories on top. A treaty in international law can be an executive agreement in domestic law, and the reverse is also possible in some jurisdictions. The definition does not resolve these domestic constitutional questions; it only establishes what counts as a treaty for purposes of international law.
Ratification and Consent to Be Bound
The Meaning of Ratification
Ratification is the international act by which a state formally expresses its consent to be bound by a treaty. It is distinct from signature. Signature may authenticate a text and create a provisional obligation not to defeat the treaty’s object and purpose, but it does not ordinarily bind the state unless the treaty provides otherwise. Ratification is the step that transforms a signed text into a binding obligation. The Vienna Convention defines ratification as the international act whereby a state establishes on the international plane its consent to be bound by a treaty.
The term “ratification” is sometimes used loosely to describe the domestic legislative approval process that precedes the international act. In the United States, the Senate “ratifies” a treaty in common parlance, but the constitutional term is “advice and consent,” and the actual international ratification is performed by the president after the Senate approves. This domestic-international confusion is widespread and reflects the layered nature of treaty-making. A state may sign a treaty, submit it to its legislature for approval, and then deposit an instrument of ratification with the depositary. Each step has distinct legal consequences, and failure at any step can prevent the treaty from binding that state.
Entry into Force
Entry into force is the moment when a treaty begins to have legal effect for the states that have consented to be bound. The Vienna Convention provides that a treaty enters into force in such manner and upon such date as it may provide or as the negotiating states may agree. Multilateral treaties typically specify a threshold number of ratifications before entry into force. The Convention on the Rights of the Child, for example, entered into force on September 2, 1990, the thirtieth day following the deposit of the twentieth instrument of ratification or accession . Bilateral treaties often enter into force upon the exchange of instruments of ratification or on a date specified in the text.
Entry into force is not the same as adoption or signature. A treaty can be adopted by a diplomatic conference, signed by dozens of states, and still not be in force for years or decades. The 1986 Vienna Convention on the Law of Treaties between States and International Organizations, for example, was adopted but has not entered into force because it has not received the required number of ratifications. For a state that consents to be bound after a treaty has already entered into force, the treaty enters into force for that state on the date of its ratification or accession, unless the treaty provides otherwise. The timing of entry into force can have significant consequences for the scope of obligations.
The State Party
A state party is a state that has consented to be bound by a treaty and for which the treaty is in force. The Vienna Convention distinguishes between a “contracting state,” which has consented to be bound but for which the treaty has not yet entered into force, and a “party,” which is bound and for which the treaty is in force. The distinction matters for obligations: a contracting state may have a provisional duty not to defeat the object and purpose of the treaty, but a state party is fully bound by all its provisions. The rights and obligations of state parties are governed by the treaty itself, supplemented by the Convention’s default rules.
The status of state party is not permanent. A state may withdraw from a treaty in accordance with its provisions or by consent of all parties. The Vienna Convention provides that a treaty which contains no provision regarding termination or withdrawal is not subject to denunciation or withdrawal unless the parties intended to admit the possibility or the right may be implied by the nature of the treaty. In practice, most multilateral treaties include withdrawal clauses, and the question of whether a state can withdraw without such a clause is contested. The status of state party also determines participation in treaty bodies, voting rights, and eligibility for dispute settlement mechanisms.
Treaty Versus Non-Binding Instrument
The Intent Test
The decisive distinction between a treaty and a non-binding instrument is intent. Did the parties intend to create legal rights and obligations under international law? This question is answered by examining the text, the context, and the subsequent conduct of the parties. The ICJ has emphasized that the label is not determinative and that the actual terms of the instrument, read in light of the circumstances of its adoption, govern. In the Somalia v. Kenya case, the Court found that a Memorandum of Understanding was binding because its language was mandatory and it stated that it would enter into force upon signature . In other cases, courts have found that instruments labeled as agreements were non-binding because their language was aspirational.
The intent test creates a gray zone that is larger than many practitioners acknowledge. Political declarations, joint statements, and memoranda of understanding often contain a mix of binding and non-binding provisions. A single instrument may create binding obligations on some matters and merely record political commitments on others. The Vienna Convention does not provide a bright-line rule for separating the two. The result is that the legal status of many international instruments is uncertain until a court or tribunal rules on the question, and many disputes are settled politically rather than judicially. This uncertainty is not a defect of the Convention; it reflects the reality that states often prefer ambiguity.
Language That Signals Bindingness
Although no single word determines treaty status, certain linguistic markers are strong indicators. Binding instruments typically use verbs such as “shall,” “agree,” “undertake,” and “shall enter into force.” They identify the parties as “states” or “governments” and refer to “rights,” “obligations,” and “provisions.” Non-binding instruments often use softer language such as “should,” “endeavor,” “intend,” and “decide.” They may refer to “participants” rather than “parties” and to “understandings” rather than “obligations.” These markers are not conclusive, but they are the first thing a practitioner looks for when assessing an instrument’s status.
The linguistic test has limits. Drafters can deliberately use ambiguous language to preserve flexibility, and courts may look beyond the text to the parties’ conduct. A state that consistently treats an instrument as binding, invokes it in diplomatic correspondence, and implements its provisions through domestic legislation may be estopped from later denying its binding character. Conversely, a state that treats an instrument as political may undermine its legal force even if the text is mandatory. The intent test is therefore not purely textual; it is contextual and behavioral. This is why treaty interpretation is often described as an art rather than a science.
The Memorandum of Understanding Problem
The Memorandum of Understanding is the most common source of confusion in treaty classification. Historically, many binding treaties were labeled MOUs, particularly among Commonwealth states. In recent decades, under the influence of practice guides and a desire for flexibility, states have increasingly used the MOU label for non-binding instruments. The result is that the same label now covers instruments that are binding treaties in simplified form and instruments that are purely political. The ICJ’s Somalia v. Kenya decision confirmed that an MOU can be a treaty, but it did not establish a rule that MOUs are presumptively binding or non-binding. The label has become nearly meaningless as a guide to legal status.
For practitioners, the MOU problem requires careful textual analysis. The key questions are whether the instrument uses mandatory language, whether it specifies entry into force, whether it creates rights and obligations, and whether the parties have treated it as binding in their subsequent conduct. The answers are not always clear. Some MOUs are deliberately drafted to be ambiguous, allowing the parties to claim binding force when convenient and deny it when not. This strategic ambiguity is a feature of modern diplomacy, not a bug. It allows states to cooperate on sensitive issues without incurring the domestic political costs of formal treaty ratification.
Reservations and the Integrity of Treaties
What a Reservation Does
A reservation is a unilateral statement made by a state when signing, ratifying, accepting, approving, or acceding to a treaty, by which it purports to exclude or modify the legal effect of certain provisions of the treaty in their application to that state. Reservations allow states to join treaties without accepting every obligation, which can increase participation but can also fragment the treaty’s obligations. The Vienna Convention permits reservations unless the treaty prohibits them, they are incompatible with the object and purpose of the treaty, or the treaty requires that all parties accept the reservation.
The compatibility of a reservation with the object and purpose of a treaty is often contested. Who decides whether a reservation is compatible? The Convention does not establish a central authority, and in practice each state party makes its own judgment. This can lead to a situation in which a reserving state considers itself a party while other states consider the reservation invalid and the reserving state not a party. The European Court of Human Rights addressed this in Bellios v. Switzerland, recharacterizing a Swiss “interpretative declaration” as a reservation and applying the permissibility test. The case illustrates that the label a state attaches to its statement does not control its legal effect.
Interpretative Declarations
An interpretative declaration is a statement by a state about its understanding of a treaty provision. Unlike a reservation, it does not purport to exclude or modify the treaty’s legal effect. But the line between a reservation and an interpretative declaration can be thin. If a state declares that it interprets a provision in a particular way and that its consent to be bound is conditioned on that interpretation, the declaration is functionally a reservation. The International Law Commission has addressed this problem in its work on reservations, and the European Court of Human Rights has applied a substantive test that looks beyond the label to the actual effect of the statement.
The distinction matters because reservations are subject to the compatibility test while interpretative declarations are not. A state that wants to influence interpretation without formally reserving may issue a declaration, but if the declaration has the effect of excluding or modifying an obligation, it may be treated as a reservation. The legal consequences are significant: an invalid reservation can prevent a state from becoming a party, while an interpretative declaration cannot. The uncertainty surrounding this distinction is one reason why treaty bodies and courts have increasingly looked to substance over form.
Invalidity and Jus Cogens
The Vienna Convention provides that a treaty is void if, at the time of its conclusion, it conflicts with a peremptory norm of general international law, known as jus cogens. A peremptory norm is a norm accepted and recognized by the international community of states as a whole as one from which no derogation is permitted and which can be modified only by a subsequent norm of the same character. The prohibition of genocide, slavery, and torture are commonly cited examples. The jus cogens doctrine is a limit on the freedom of states to make treaties: they cannot contract out of fundamental norms.
The invalidity of a treaty on jus cogens grounds is rare in practice. The Convention also provides for invalidity based on error, fraud, corruption, coercion of a representative, and coercion of a state by threat or use of force. These grounds are narrowly defined and rarely invoked successfully. The existence of these rules reflects the idea that treaty-making is not an unconstrained sovereign act; it operates within a legal framework that includes limits. But the enforcement of these limits depends on states bringing claims, and the political costs of alleging that another state’s treaty is void on jus cogens grounds are high. The doctrine is therefore more important as a statement of values than as a frequent basis for invalidating agreements.
Why the Definition Matters
Domestic Constitutional Consequences
The international law definition of a treaty has significant domestic constitutional consequences, but they vary by jurisdiction. In the United States, the distinction between a treaty and an executive agreement determines whether the Senate must approve the agreement by a two-thirds vote. In the Philippines, the Supreme Court has addressed whether an international agreement can be concluded as an executive agreement without Senate concurrence, holding that the determination depends on the parties’ intent and the agreement’s subject matter . The international law definition sets the outer boundary of what counts as a treaty, but domestic law determines the procedures required for ratification and implementation.
This layered system creates opportunities for strategic behavior. A government that wants to avoid legislative approval may choose an executive agreement, even for a significant commitment, knowing that the agreement will still be a treaty under international law. The domestic political advantage comes at the cost of durability: executive agreements can often be terminated more easily than treaties. The choice of instrument is therefore not merely a technical legal question but a political one, reflecting calculations about domestic support, international credibility, and the desired permanence of the commitment.
Pacta Sunt Servanda and the Rule of Law
The foundational principle of treaty law is pacta sunt servanda: agreements are binding and must be performed in good faith. The Vienna Convention states that every treaty in force is binding upon the parties to it and must be performed by them in good faith. This principle is the reason treaties are considered a source of international law and a foundation of the international legal order. Without it, the entire edifice of treaty-making would collapse into a system of political promises. The definition of a treaty is therefore not just a taxonomic exercise; it identifies the instruments that are backed by this principle.
The rule of law in international affairs depends on the ability to distinguish binding commitments from political rhetoric. If everything a state says or signs could be a treaty, the concept would lose its meaning. If nothing is a treaty until a court says so, the concept would be too uncertain to guide behavior. The Vienna Convention definition attempts to strike a balance: it provides objective criteria (written form, states, governed by international law) while leaving room for contextual judgment about intent. The balance is imperfect, but it is the best available framework for determining what counts as a treaty in international law.
What a treaty is, in the end, is a promise that the law takes seriously. The definition matters because it determines when that seriousness attaches. A state that signs an instrument knowing it will be bound has made a choice to constrain its future freedom. A state that signs a non-binding instrument has made a different choice. The line between the two is sometimes blurred, sometimes manipulated, and sometimes litigated. But it is a line that international law insists on drawing, because without it, the distinction between law and politics would disappear. The Vienna Convention’s definition is the starting point for drawing that line, even if it is not the final word.