Quantitation of mutans streptococci (MS) and Lactobacillus has been proposed as a method for the identification of patients at high risk for dental caries. The purpose of the present study was to investigate the analytical and physiological variability of this procedure. Variability due to specimen collection and processing was investigated by repeated sampling studies. Sample stability was investigated over storage periods of up to 72 hr at 5 degrees C, room temperature, and 37 degrees C. Physiological variability was investigated by performance of serial analyses on ten individuals, who collected six samples at intervals on a single day, and rising and noon samples on ten subsequent days. Sample collection, rather than sample processing, was found to be the major factor determining the imprecision of salivary microbial analysis in the majority of cases. However, individual subjects varied considerably in the consistency with which they provided saliva samples. Imprecision due to sample processing was relatively small, with coefficients of variations of 2.3% for MS counts and 2.1% for Lactobacillus counts. Samples were generally stable over a wide temperature range for at least 72 hours, although there was some loss of viability of lactobacilli on prolonged storage at room temperature or 37 degrees C in one experiment. Rising samples yielded higher counts than samples collected after breakfast and toothbrushing. Day-to-day variability was considerable, with 95% confidence limits exceeding 1 log in 28% of data sets for MS count and 39% of data sets for Lactobacillus.