In patients, urinary levels of pentamidine have been shown to reflect pulmonary deposition of aerosolized drug. Using urinary levels and air filter samples, we assessed factors responsible for health care worker (HCW) exposure. We measured serial urine samples in HCWs who administered aerosol pentamidine over an 11-month period and compared them with serial urine levels measured over 30 days in a normal volunteer in whose lungs a known amount of pentamidine (3.39 mg) had been deposited. Ambient exposure to pentamidine was determined by continuous high volume air sampling in the treatment room during routine therapy. In addition, the amount of pentamidine released by six HIV-positive subjects, performing tidal breathing with a Respirgard II nebulizer in an airtight booth, was measured by extracting air from the booth through a filter. The effect of adding noseclips, of coughing (with nebulizer shut down), and of removing the nebulizer from the patient's mouth without turning it off, were determined. Pentamidine in the urine of the normal volunteer reached a peak concentration of 9.5 ng/mg creatinine/ml and was detectable for 30 days following the exposure. In HCWs, pentamidine was detected intermittently in four of five individuals with levels as high as 18.2 ng/mg creatinine/ml. Samples of ambient treatment room air indicated small daily releases of pentamidine (0.013 +/- 0.02 mg per patient treated), but simultaneous urine levels in HCWs were negative. The data from the airtight booth revealed that removing the nebulizer from a patient's mouth without turning it off caused a 360-fold increased in pentamidine release compared to tidal breathing. Coughing resulted in a 6.9 (range 0.9-14.2)-fold increase in release, while the addition of noseclips had no significant effect. The pattern of intermittently positive urine tests and the low levels of ambient pentamidine detected in the air of the treatment room suggest that HCWs are being exposed to episodic but high concentrations of pentamidine. High level exposure is most likely to occur during treatment interruptions which are usually precipitated by coughing episodes. Because of the intermittent pattern of exposure and slow clearance of pentamidine, urine assay is useful for detecting high intermittent exposure. Random air sampling is a sensitive indicator of low level exposures but may not detect episodic high level releases.