To detect whether there was geographic clustering of Pneumocystis carinii pneumonia cases among patients with human immunodeficiency virus (HIV) infection, we performed a retrospective analysis of a clinical database. The rates of pneumocystosis were analyzed by zip code zones for evidence of geographical clustering. During the study period, 118 patients at our AIDS Treatment Center had a first episode of P. carinii pneumonia. An analysis of the 24 zip code zones for which a P. carinii pneumonia rate was calculated (requiring a denominator of at least 10 known HIV- infected individuals residing in that zone) showed a trend toward geographic clustering (p = 0.07); when all 45 Cincinnati zip code zones were included in the analysis, clustering of cases was observed (p = 0. 02). By contrast, no clustering was observed for 52 HIV-infected control subjects with respiratory disease or for 960 HIV-infected patients treated at our center during the same time period. These data raise intriguing questions about exposure to exogenous sources of P. carinii and suggest the need for prospective studies.