This article examines the admissibility of forensic science and medicine in criminal proceedings. In Part ii, we explain how reliability-based admissibility standards in the United States have been unevenly applied to expert evidence in civil and criminal cases and have not prevented wrongful convictions. In Part iii, we review a recent Consultation Paper (and report) issued by the Law Commission of England and Wales. Though focused on the need for ‘sufficiently reliable’ expert opinion evidence, we challenge its contemplation of easier admissibility for experience-based forensic sciences and techniques traditionally admitted. In Part iv we examine the evolving law on the admissibility of expert evidence in Canada. In response, we argue that while front-end reforms to the organization and practice of forensic science and medicine, advocated by the Goudge Inquiry and the American National Academy of Sciences, appear more promising than reliance on the adversary system, the gate-keeping role of trial judges should be strengthened. In the concluding section, we contend that threshold reliability standards should be grounded in criminal-justice system values, emerging empirical insights about the weakness of the adversarial trial and be sensitive to the particular evidence and its use, rather than applied mechanically using simplistic models of science and expertise.