Fecal specimens from several laboratory animal species were tested for rotavirus antigen by Rotazyme II, a commercially available enzyme-linked immunosorbent assay (ELISA) that is widely used in human diagnostic studies. Fecal samples from rabbits, hamsters, guinea pigs, dogs, and cats tested negative; whereas those from rats and mice yielded a high proportion of positive results. Rats had the highest rate with 82% of the samples being positive. However, the presence of rotavirus in positive rodent samples could not be confirmed by virus isolation, electron microscopy or blocking ELISA using anti-EDIM mouse rotavirus serum. Several lines of evidence indicated that these positive reactions were false positives, apparently due to a non-specifically reacting substance in the diet of rats and mice. All the positive fecal samples were from rats and mice that had been fed nonautoclaved diet. Samples from rodents fed autoclaved diet were consistently negative in the Rotazyme test. When rats fed autoclaved diet were subsequently fed nonautoclaved diet, their stool converted from negative to positive within 6 hours. Conversely, rats with positive stool samples converted to negative within 15 hours when fed autoclaved diet. Similar results were found with mice. Positive fecal specimens and nonautoclaved rodent diet both contained a substance that apparently attached nonspecifically to the antibody coated beads used in the ELISA and reacted directly with the substrate in the absence of the conjugate. This substance was heat labile and trypsin sensitive, suggesting that it was a protein.