To improve the detection of Mycoplasma pulmonis contamination of isolates of cilia-associated respiratory (CAR) bacillus, we developed a nested PCR method using primers for 16S rRNA gene sequences. Of 140 samples of 16 different CAR bacillus isolates, 73 (52%) were inhibitory in the first PCR, as indicated by the absence of amplicons of the internal control, but only 11 of 140 (7.9%) were inhibitory in the second PCR. Of 27 samples known to contain M. pulmonis, only 12 (44%) were positive in the first PCR, but 25 of 27 (93%) were positive in the second PCR. Nested PCR also detected M. pulmonis in 21 of 61 (34%) CAR bacillus samples from which M. pulmonis could not be cultured and identified 2 additional M. pulmonis-contaminated CAR bacillus isolates. Of 359 respiratory and reproductive tract lavage samples from rats and mice, 35 (9.8%) were inhibitory in the first PCR, but only 15 (4.2%) were inhibitory in the second PCR. Of 72 lavage specimens from rats inoculated with an avirulent, poorly infective M. pulmonis strain, 14 (19%) were positive by nested PCR, but only 2 of 72 (2.8%) were positive by culture. Nested PCR also detected M. pulmonis in 14 of 20 (70%) paraffin sections of lung and trachea from rats and mice inoculated with CAR bacillus isolates known to contain M. pulmonis, whereas single PCR gave no positive results. We conclude that nested PCR is superior to single PCR or culture for detecting M. pulmonis, and that M. pulmonis is present in all but four CAR bacillus isolates in our collection that were from naturally infected rats; the four isolates that were exceptions were obtained from rats from a single colony.