To evaluate the effect of diet on results obtained by use of 2 commercial test kits for detection of occult blood in feces, 5 dogs were fed 7 diets in randomized sequence. Dry and canned diets with various principal ingredients were evaluated. Each diet was offered twice over a 24-hour period, followed by a 36-hour nonfeeding period. Fecal specimens were collected twice daily, and tests for occult blood were performed within 12 hours. The dietary origin of fecal specimens was confirmed by use of colored markers fed with each diet, and was correlated with estimates of gastrointestinal tract transit time. A modified guaiac paper test and an ortho-tolidine tablet test were performed on each specimen. Of 59 specimens, 4 were positive for occult blood, using the ortho-tolidine tablet test. Three positive results were associated with a mutton-based canned diet, and 1 positive result was associated with a canned beef-based diet. Of 59 specimens, 11 were positive for occult blood, using the modified guaiac paper test. Four positive results were associated with the mutton diet, and 3 positive results were associated with the beef diet. Of the remaining 5 diets, 4 caused 1 positive reaction. Results were inconsistent with the null hypothesis that the distribution of positive occult blood test results is not affected by diet (P < 0.025), and indicate that diet can affect the specificity of peroxidase-based tests for detection of occult blood in canine feces. Diet modification prior to testing is recommended.