The numbers and types of microorganisms contaminating a commercially available laboratory animal diet were examined by plate-count methods at all stages of production. During the cooking and pelleting stages of manufacture there was a marked reduction in numbers of heat-labile vegetative contaminants introduced via the raw materials, and at subsequent stages plate counts at 37 and 25 degrees C corresponded closely to counts of viable aerobic bacterial endospores. It would seem that the count on the pelleted diet was determined principally by the numbers of bacterial spores being introduced with the ingredients. The response to gamma-irradiation of the innate microflora contaminating the pelleted diet was characteristic of that generally seen with populations of aerobic bacterial endospores.