To assess unfiltered drinking water as a source of cryptosporidium infection in patients with the acquired immunodeficiency syndrome (AIDS) the prevalence of cryptosporidiosis among persons with AIDS in Los Angeles County was assessed by water service area. One water distributor, serving approximately 60% of the county's residents (area B), has consistently employed filtration. The other company, which serves the remainder of the county (area A), did not institute filtration until mid-December 1986. This difference provided a 'natural experiment' in which to assess the effect of municipal water filtration on the level of cryptosporidiosis among persons with AIDS. The prevalence of cryptosporidiosis among AIDS patients was compared for the two water service areas for the time period (1983-6) preceding the implementation of filtration in area A. From 1983 to 1986 the age-standardized prevalence of cryptosporidiosis among AIDS patients was 32% lower in area A (4.2%), which received unfiltered water, than in area B (6.2%). Following addition of filtration in area A, the prevalence of cryptosporidiosis among AIDS patients decreased by 20%; however, a decline, of 47%, was also observed in area B. The similar baseline levels of cryptosporidiosis and the corresponding post-filtration decline in both areas suggest that filtration had no effect on levels of cryptosporidiosis among persons with AIDS. Thus it does not appear that municipal drinking water is an important risk factor for cryptosporidiosis in AIDS patients residing in Los Angeles County.