A case-control study of spontaneous diabetes mellitus in dogs was undertaken, using 2 veterinary data bases. The Veterinary Medical Data Program (VMDP) contained records of 1,019 cases of canine diabetes from 14 university-affiliated veterinary hospitals. The Animal Medical Center (AMC), a private veterinary hospital which has not participated in the VMDP, contained records of 449 diabetes cases. Each data base was analyzed separately, control groups being chosen from all admissions, excluding diabetic animals. Summary odds ratios by sex adjusted for age and breed indicated significantly (P less than 0.05) elevated risks for entire females and neutered females compared with that for entire males. The VMDP data indicated a significantly elevated risk for castrated males, whereas the risk derived from AMC data was not significantly different from 1. Analysis of risks by breed adjusting for age and sex identified Poodles as being at significantly excess risk, and German Shepherd Dogs, Cocker Spaniels, Collies, and Boxers at significantly decreased risk in both data sets. The male-female risk ratio changed with age from 1 at less than 1 year of age to a predominance of females at older ages. In the AMC data base, diabetes was significantly associated with cataracts in dogs of both sexes combined. Diabetes was also significantly associated with benign mammary tumors in female dogs.