The therapeutic efficacy of nebulised pentamidine in the prophylaxis of Pneumocystis carinii pneumonia (PCP) depends on the absolute pulmonary deposition of the drug. We studied the performance of a new nebuliser (Pentasave) by comparison both in vitro and in vivo with a standard nebuliser (Respirgard II). In vitro, deposition of pentamidine labelled with technetium-99m human serum albumin was measured indirectly by capturing inhaled particles on an absolute filter and measuring radioactivity with a gamma camera. The nebulisers were initially assessed with a pentamidine dose of 100 mg in 5 ml at 44 psi and an air flow of 10 l/min for Respirgard II and 16 l/min for Pentasave. Nebuliser output, expressed as the percentage of the initial nebuliser radioactivity captured by the inhalation filter, was 15% +/- 2% (mean +/- SD) for Respirgard II, and significantly increased to 23% +/- 3% for an initial version and to 33% +/- 2% for the final version of Pentasave. Measurements with a gamma camera in a group of ten patients with human immunodeficiency virus infection were made in vivo. The results revealed that pulmonary drug distributions are good using both Respirgard II and Pentasave. The literature reports that once-monthly pulmonary deposition of 9 mg pentamidine seems enough to produce prophylactic effects against Pneumocystis carinii. We measured pulmonary pentamidine deposition of 20.22 +/- 4.31 mg (mean +/- SD) using Respirgard II (with 300 mg in 5 ml) and of 16.00 +/- 7.18 mg using Pentasave (with 150 mg in 6 ml). These findings show that the therapeutic dose of pentamidine (9 mg) was widely exceeded with both nebulisers. Further investigations might demonstrate that about 200 mg and 125 mg pentamidine for Respirgard II and Pentasave, respectively, will achieve a pulmonary deposition of therapeutic dose, allowing significant savings in terms of drug and expense.