This article reviews research on the Minimalist instructional design model, a learner-centred approach to the design of instructional materials such as computer program manuals or on-line help. Studies in this paradigm have typically compared minimalist materials against traditional "system-centred" materials. Additionally, some studies have investigated the effects of specific minimalist design heuristics. However, small samples in many original studies have meant effect size estimates are imprecise. Meta-analytic methods were applied to 13 experimental effect sizes comparing the minimalist and system-centred approaches, and 6 effects investigating the effects of specific minimalist heuristics ("support error recognition and recovery" and "slash the verbiage"). The overall mean effect on learning in favour of the minimalist approach was large (d = 1.12), broadly supporting the minimalist approach compared to system-centred designs. The mean effect for tests of the "support error recognition and recovery" principle was moderate (d = 0.59), and the mean effect for tests of the "slash the verbiage" principle was large (d = 0.89). These results indicate the application of minimalist design principles to self-study manuals for learning to use computer programmes is associated with more effective learning. By applying these principles, designers of such manuals can assist learners, as well as gaining a competitive edge. (Contains 3 tables.)