Aims To quantify the extent to which ethnic differences in muscular strength might account for the substantially higher prevalence of diabetes in black and South‐Asian compared with white European adults. Methods This cross‐sectional study used baseline data from the UK Biobank study on 418 656 white European, black and South‐Asian participants, aged 40–69 years, who had complete data on diabetes status and hand‐grip strength. Associations between hand‐grip strength and diabetes were assessed using logistic regression and were adjusted for potential confounding factors. Results Lower grip strength was associated with higher prevalence of diabetes, independent of confounding factors, across all ethnicities in both men and women. Diabetes prevalence was approximately three‐ to fourfold higher in South‐Asian and two‐ to threefold higher in black participants compared with white European participants across all levels of grip strength, but grip strength in South‐Asian men and women was ~ 5–6 kg lower than in the other ethnic groups. Thus, the attributable risk for diabetes associated with low grip strength was substantially higher in South‐Asian participants (3.9 and 4.2 cases per 100 men and women, respectively) than in white participants (2.0 and 0.6 cases per 100 men and women, respectively). Attributable risk associated with low grip strength was also high in black men (4.3 cases) but not in black women (0.4 cases). Conclusions Low strength is associated with a disproportionately large number of diabetes cases in South‐Asian men and women and in black men. Trials are needed to determine whether interventions to improve strength in these groups could help reduce ethnic inequalities in diabetes prevalence. What's new? In participants in the UK Biobank study, grip strength was significantly associated with odds of diabetes, independent of major confounding factors in all ethnic groups. Grip strength was lower (by ˜ 5–6 kg) in South‐Asian men and women compared with black and white European men and women. Lower grip strength combined with higher diabetes prevalence resulted in the attributable risk for diabetes associated with low grip strength being substantially higher in South‐Asian people (3.9 and 4.2 cases per 100 men and women) than in white people (2.0 and 0.6 cases). Attributable risk was also high in black men (4.3 cases) but not in black women (0.4 cases).