The risk of type 2 diabetes among obese adults who are metabolically healthy has not been established. We systematically searched Medline (1946–August 2013) and Embase (1947–August 2013) for prospective studies of type 2 diabetes incidence (defined by blood glucose levels or self-report) among metabolically healthy obese adults (defined by body mass index [BMI] and normal cardiometabolic clustering, insulin profile or risk score) aged ≥18 years at baseline. We supplemented the analysis with an original effect estimate from the English Longitudinal Study of Ageing (ELSA), with metabolically healthy obesity defined as BMI ≥ 30 kg m−2 and <2 of hypertension, impaired glycaemic control, systemic inflammation, adverse high-density lipoprotein cholesterol and adverse triglycerides. Estimates from seven published studies and ELSA were pooled using random effects meta-analyses (1,770 healthy obese participants; 98 type 2 diabetes cases). The pooled adjusted relative risk (RR) for incident type 2 diabetes was 4.03 (95% confidence interval = 2.66–6.09) in healthy obese adults and 8.93 (6.86–11.62) in unhealthy obese compared with healthy normal-weight adults. Although there was between-study heterogeneity in the size of effects (I2 = 49.8%; P = 0.03), RR for healthy obesity exceeded one in every study, indicating a consistently increased risk across study populations. Metabolically healthy obese adults show a substantially increased risk of developing type 2 diabetes compared with metabolically healthy normal-weight adults. Prospective evidence does not indicate that healthy obesity is a harmless condition.