This study was undertaken to examine quantitatively the risks to human health posed by heterotrophic plate count (HPC) bacteria found naturally in ambient and potable waters. There is no clear-cut evidence that the HPC bacteria as a whole pose a public health risk. Only certain members are opportunistic pathogens. Using the four-tiered approach for risk assessment from the National Academy of Sciences, hazard identification, dose-response modeling, and exposure through ingestion of drinking water were evaluated to develop a risk characterization, which estimates the probability of infection for individuals consuming various levels of specific HPC bacteria. HPC bacteria in drinking water often include isolates from the following genera: Pseudomonas, Acinetobacter, Moraxella, Aeromonas, and Xanthomonas. Other bacteria that are commonly found are Legionella and Mycobacterium. All these genera contain species that are opportunistic pathogens which may cause serious diseases. For example, the three nonfermentative gram-negative rods most frequently isolated in the clinical laboratory are (1) Pseudomonas aeruginosa, (2) Acinetobacter, and (3) Xanthomonas maltophilia. P. aeruginosa is a major cause of hospital-acquired infections with a high mortality rate. Aeromonas is sometimes associated with wound infections and suspected to be a causative agent of diarrhea. Legionella pneumophila causes 4%-20% of cases of community-acquired pneumonia and has been ranked as the second or third most frequent cause of pneumonia requiring hospitalization. The number of cases of pulmonary disease associated with Mycobacterium avian is rapidly increasing and is approaching the incidence of M. tuberculosis in some areas. Moraxella can cause infections of the eye and upper respiratory tract. The oral infectious doses are as follows in animal and human test subjects: P. aeruginosa, 10(8)-10(9); A, hydrophila, > 10(10); M. avium, 10(4)-10(7); and X. maltophilia, 10(6)-10(9). The infectious dose for an opportunistic pathogen is lower for immunocompromised subjects or those on antibiotic treatment. These bacteria have been found in drinking water at the following frequencies: P. aeruginosa, < 1%-24%; Acinetobacter, 5%-38%; X. maltophilia, < 1%-2%; Aeromonas, 1%-27%; Moraxella, 10%-80%; M. avium, < 1%-50%; and L. pneumophila, 3%-33%. These data suggest that drinking water could be a source of infection for some of these bacteria. The risk characterization showed that risks of infection from oral ingestion ranged from a low of 7.3 x 10(-9) (7.3/billion) for low exposures to Aeromonas to higher risks predicted at high levels of exposure to Pseudomonas of 9 x 10(-2) (98/100). This higher risk was only predicted for individuals on antibiotics. Overall, the evidence suggests that specific members of HPC bacteria found in drinking water may be causative agents of both hospital- and community-acquired infections. However, the case numbers may be very low and the risks represent levels generally less than 1/10,000 for a single exposure to the bacterial agent. Future research needs include (1) determining the seasonal concentrations of these bacteria in drinking water, (2) conducting adequate dose-response studies in animal subjects or human volunteers, (3) determining the health risks for an individual with multiple exposures to the opportunistic pathogens, and (4) evaluating the increase in host susceptibility conferred by antibiotic use or immunosuppression.