This paper explores the progress of educational programme evaluation over three decades through the lens of a group of evaluators primarily from the UK and the US who met periodically from 1972 to 2004 to review the state of the art of educational programme evaluation--what it could and could not do--in relation to the complexity of programme initiatives and the policy and political contexts of the times. These meetings came to be known as the Cambridge conferences as they were always held at a university college in Cambridge. They were sponsored in the main by the Nuffield Foundation and, in 2004, by the ESRC. The nucleus of original group membership was retained over the years for continuity with additional members joining as the specific focus of the conference changed and new ideas were sought. There have been six conferences. This paper first locates the contextual origin of and reason for the first conference and the manifesto that was written from it. Secondly it documents the revisiting of that manifesto at the sixth conference leading to the construction of a "new" manifesto that reflects changes required in evaluation methodology in the socio/political context of the first decade of the twenty-first century.