Between June 16 and October 9, 1974, 9 neonates at a small, community hospital were stricken with an unusual, serious illness manifested by peritonitis and pneumoperitoneum; 3 died. Although the illness was initially thought to be necrotizing entercolitis, clinical, laboratory, and epidemiologic evidence strongly suggested that it was instead the result of gastrointestinal perforation. in case-control studies employing 3 different conposure to a particular nurses' aide. Other studies including a comparison of expected and actual exposures of ill infants to nursery personnel further linked this nurses' aide to illness. Since rectal temperature-taking was the only procedure possibly predisposing to gastrointestinal perforation that was routinely practiced in the nursery, it was hypothesized that the illness might be the result of rectal perforations. In order that rectal temperature-taking technique could be observed, each nurse and nurses' aide on the OB-GYN service was asked to take part in a general practical examination of nursing skills on a life-like baby doll. The mean and median depths to which nursing personnel inserted the thermometer exceeded the maximum depth recommended to prevent perforation. The nurses' aide epidemiologically associated with illness inserted the thermometer to almost twice the maximum recommended depth-farther than all the personnel who worked primarily in the nursery. After this nurses' aide was removed from the nursery and axillary temperature-taking replaced rectal temperature-taking as the nursery routine, the outbreak ceased.