Peace Agreement Amnesties Dataset

This dataset uses the PA-X Peace Agreement database created by PeaceRep to identify peace agreements with amnesty provisions. PA-X defines peace agreements as ‘formal, publicly-available documents, produced after discussion with conflict protagonists and mutually agreed to by some or all of them, addressing conflict with a view to ending it’. In March 2025, it contained data on 2055 peace agreements found in more than 150 peace processes, relating to 106 states around the world between 1990 and 2024. In the PA-X database, each of these agreements is coded in relation to a range of factors relating to the context of the agreements and their substantive content. This includes a variable called ‘Amnesty/pardon proper’ that reproduces any provisions of a peace agreement that provide for ‘amnesties or pardons, staying or stopping prosecution, and their modalities (may include provisions of prohibition of amnesty)’. PA-X codes 205 peace agreements as having an ‘amnesty/pardon proper’. For the Peace Agreement Amnesty dataset, the PA-X data was recoded using an Excel spreadsheet. In doing so, cases that were coded as providing for an amnesty in PA-X were excluded from this dataset where (1) they related to a localized peace agreement; (2) the provisions related to domestic rules on amnesty rather than creating a commitment to grant amnesty; or (3) the text was ‘unilateral’ rather than a negotiated agreement. The dataset also exclude peace agreements that restated a commitment to amnesty that was made in an earlier agreement. The sample includes some agreements that were ‘unsigned’. As a result of this filtering process, the resulting Peace Agreement Amnesty dataset documents 117 peace agreements providing amnesty in relation to internal armed conflicts from 1990 to 2024 in 47 countries.