School-based curriculum innovations, including those in science education, are usually not adequately evaluated, if at all. Furthermore, current procedures and instruments for programme evaluations are often unable to support evidence-based decision-making. We suggest that adopting fidelity of implementation (FOI) criteria from healthcare research can both characterize and narrow the separation between programme intent and actual implementation, which is a mandatory stage of evaluation before determining overall programme value. We demonstrate how such a process could be applied by science educators using data from a secondary school in Singapore that had devised a new curriculum to promote interest, investigative processes, and knowledge in science. Results showed that there were ambivalent student responses to this programme, while there were high levels of science process skill instruction and close alignment with the intended lesson design. The implementation of this programme appeared to have a satisfactory overall level of FOI, but we also detected tensions between programme intent and everyday classroom teaching. If we want to advance science education, then our argument is that applying FOI criteria is necessary when evaluating all curricular innovations, not just those that originate from schools.