Induced pluripotent stem cell (iPSC) technologies have provided in vitro models of inaccessible human cell types, yielding new insights into disease mechanisms especially for neurological disorders. However, without due consideration, the thousands of new human iPSC lines generated in the past decade will inevitably affect the reproducibility of iPSC-based experiments. Differences between donor individuals, genetic stability and experimental variability contribute to iPSC model variation by impacting differentiation potency, cellular heterogeneity, morphology, and transcript and protein abundance. Such effects will confound reproducible disease modelling in the absence of appropriate strategies. In this Review, we explore the causes and effects of iPSC heterogeneity, and propose approaches to detect and account for experimental variation between studies, or even exploit it for deeper biological insight.