PurposeThe importance of dietary diversification for type 2 diabetes (T2D) risk remains controversial. We investigated associations of between- and within-food group variety with T2D, and the role of dietary diversification for the relationships between previously identified dietary patterns (DPs) and T2D among Ghanaian adults.MethodsIn the multi-center cross-sectional Research on Obesity and Diabetes among African Migrants (RODAM) Study (n = 3810; Ghanaian residence, 56%; mean age, 46.2 years; women, 63%), we constructed the Food Variety Score (FVS; 0–20 points), the Dietary Diversity Score (DDS; 0–7 points), and the Diet Quality Index-International (DQI-I) variety component (0–20 points). The associations of these scores, of a “rice, pasta, meat and fish” DP, of a “mixed” DP, and of a “roots, tubers and plantain” DP with T2D were calculated by logistic regression.ResultsThe FVS was inversely associated with T2D, adjusted for socio-demographic, lifestyle, and anthropometric factors [odds ratio (OR) for T2D per 1 standard deviation (SD) increase: 0.81; 95% confidence interval (CI) 0.71–0.93]. The DDS and the DQI-I variety component were not associated with T2D. There was no association of the “mixed” DP and the “roots, tubers and plantain” DP with T2D. Yet, the “rice, pasta, meat and fish” DP is inversely associated with T2D (OR for T2D per 1 SD increase: 0.82; 95% CI 0.71–0.95); this effect was slightly attenuated by the FVS.ConclusionsIn this Ghanaian population, between-food group variety may exert beneficial effects on glucose metabolism and partially explains the inverse association of the “rice, pasta, meat and fish” DP with T2D.Electronic supplementary materialThe online version of this article (doi:10.1007/s00394-017-1538-4) contains supplementary material, which is available to authorized users.