Our capacity to learn from research is precisely our capacity to synthesise the findings of research to inform our actions in particular educational situations. The realisation of such a goal requires a reconceptualisation of research synthesis as partial, purposeful, situated and critical. Mathematics classroom research undertaken as part of the Learner’s Perspective Study (LPS) provides the context for this discussion. Specifically, in this paper, I am interested in establishing under what conditions we can synthesise across different parallel analyses of data generated from the same data source if those analyses are grounded in different theories. My examples employ Variation Theory, Interactional Analysis, Conversation Analysis and the analysis of Discursive Practice applied to a common body of classroom data from ten countries. Each theory brings with it a vocabulary that privileges certain constructs and downplays or ignores others. I suggest that it is not the compatibility of the theories that should be considered, but of the interpretive accounts generated by their application to a common representational record of classroom events. So, the relevant question is “Under what conditions are the interpretive accounts compatible?” This contingent compatibility focuses our attention on the use of theories as interpretive tools.