The retail food environment is a potential population-level determinant of diet and nutrition-related chronic diseases, yet little is known about its composition and association with diabetes in low- and middle-income countries. Our objectives were: (1) to describe changes in the composition of the retail food environment in Mexican neighborhoods from 2010 to 2016 and (2) to examine the association between these changes and diabetes cases diagnosed over the same period. Individual level data came from the 2016 Mexican Health and Nutrition Survey (N = 2808 adults). Neighborhood level retail food environment data for 2010 and 2016 came from the National Directory of Economic Units of Mexico. Multilevel logistic regression was used to examine the adjusted association between changes in the neighborhood density per km2 of fruit and vegetable stores, chain convenience stores and supermarkets with diabetes. Small store formats still predominate in Mexico's food environment, however there is evidence of fast increase in chain convenience stores and supermarkets. Adults living in neighborhoods that saw a decline in fruit and vegetable store density and a simultaneous increase in chain convenience store density experienced higher odds of diabetes, compared to adults who lived in neighborhoods where fruit and vegetable and convenience stores did not change (OR 3.90, 95% CI 1.61, 9.48). Considering the complex interplay between store types, understanding the mechanisms and confirming the causal implications of these findings could inform policies that improve the quality of food environments in cities.