Histologic and dietary prognostic factors for survival following naturally occurring breast cancer were studied for 145 pet dogs. Information was collected from the dog's owner and veterinarian regarding medical and reproductive history, nutritional status, treatment, tumor recurrence, and length of survival. The usual intake of all dog and table foods consumed 1 year prior to diagnosis was obtained using a validated quantitative food frequency questionnaire. A histologic malignancy score was derived based on 7 histopathologic criteria. The mean age of the dogs was 10.4 +/- 2.5 years; 37% had been ovariohysterectomized prior to diagnosis. Product-limit estimates of survival indicated that 6 factors, namely body conformation 1 year prior to diagnosis (p = 0.03), histologic tumor type (p = 0.004), histologic malignancy score (p = 0.02), histologic invasion (p = 0.002), tumor recurrence (p less than 0.0001), and completeness of surgery (p = 0.01) were of prognostic significance. In addition, when dogs were characterized by the percent of total calories they derived from fat and protein, the median survival time for dogs in the low fat group (less than 39%) with protein greater than 27%, 23-27%, and less than 23% was 3 years, 1.2 years, and 6 months, respectively (p = 0.008). For dogs in the high fat group (greater than or equal to 39%), there was no difference in survival for the different intake levels of dietary protein (p = 0.84). When these data were fitted to a proportional hazards model, recurrence, histologic score, tumor type, percent of calories derived from protein, fat group, and a protein-fat group interaction term were statistically significant. Predicted 1 year survival for dogs on a low fat diet with 15%, 25%, and 35% of total calories derived from protein was 17%, 69%, and 93%, respectively.