The conclusions from epidemiological studies are controversial between apple and pear consumption and type 2 diabetes mellitus (T2DM) risk. The present study aimed to investigate whether apple and pear consumption was inversely associated with T2DM risk, and to evaluate the potential dose–response relationship. The Cochrane library, Embase and PubMed databases were searched up to Nov 2016. Prospective cohort studies, which reported the association of apple and pear consumption with incidence of T2DM, were included. Multivariate-adjusted relative risks (RRs) for the highest versus lowest category were combined by using a random-effects model. A restricted cubic spline regression model was performed to examine the dose–response relationship. A total of 5 independent prospective cohort studies were included (14 120 T2DM incident cases and 228 315 participants). The summary estimate showed that consumption of apples and pears was associated with 18% reduction in T2DM risk (95% confidence interval (CI): 0.75, 0.88; I² = 0.00%). Dose–response analysis showed that one serving per week increment of apple and pear consumption was associated with a 3% (95% CI: 0.96, 0.98; p for trend <0.001) reduction in T2DM risk. The present meta-analysis provides significant evidence of an inverse association between apple and pear consumption and T2DM risk.