From 1 April to 31 May 1993, 64 cases of cryptosporidiosis were diagnosed within one district health authority. Forty were classified as primary cases, 35 of whom were clustered in an area supplied by a discrete public water supply that supplied the majority of homes in a large town. Most of the water in this supply is abstracted from boreholes and some is filtered before distribution. Households that received mains water from this supply were 15 times more likely to be affected than households nearby that received water from other sources. A case control study demonstrated a dose response relationship between consumption of water obtained from the town supply and risk of illness. Very low concentrations of cryptosporidial oocysts were detected in the water supply on four occasions several weeks after the outbreak. Environmental investigation failed to reveal a likely mechanism for contamination of the water supply.