'Large Scale Research Infrastructures' (LSRIs), a subcategory of megaprojects which incorporate a characteristic of a high or 'super high' level of technological uncertainty, are often undertaken as cooperative projects with long lead times by one or more national governments. A lack of research into the effect of the LSRI project's lifecycle on the research organisation is apparent, particularly when scientists and engineers exercise freedom to organise the project directly. Two case studies used senior leadership selection as proxy for the LSRI project lifecycle using a contingency theory framework. These were the Tevatron (Fermi National Accelerator Laboratory, USA) and the Large Hadron Collider at the European Organisation for Nuclear Research (CERN). This LSRI lifecycle is mapped onto lifecycles used in theory and in policy. Previous research did not detect that these projects become institutionalised, so influencing the selection of new research organisation senior leadership according to its needs at that stage of its lifecycle. This represents something of a novelty as most contingency theory work is theoretical with few attempts to use it as a conceptual framework for empirical evidence. The findings indicate a second new understanding, that while the literature characterises a leadership style transition from democratic to authoritarian as the project progresses, LSRIs exhibit a reverse transformation, probably as a product of the characteristically high level of organisational technical competence. The construction of LSRIs maps better onto the traditional project lifecycle and the National Science Foundation's large facility lifecycle than onto other lifecycles. There is a policy opportunity to commission a 'generational survey' upon the completion of an LSRI, to understand the characteristics of the 'next big machine'. The author would like to thank the CERN Archives Service (Geneva, Switzerland) for granting access to restricted sections of the CERN Archives, and also the Fermilab History and Archives Project (Batavia, IL, USA) for access to the Fermilab Archives, and both for assisting with the creation of interviewee pools. This version of the paper benefitted from the advice and comments of the SPRU Working Papers Series (SWPS) Editors and Reviewers including Katherine Lovell. The original research could not have been possible without the advice and support of my PhD supervisors: Ben Martin, the late Puay Tang, and Martin Meyer. Thanks also to my PhD viva examiners, Josh Siepel and Andy Davies, for their valuable comments and advice. This research was supported by a Graduate Teaching Assistant (GTA) grant from the University of Sussex.