With the goal of achieving economy in infection control surveillance we tested the relative efficiency of the various surveillance sources for detecting nosocomial infections (NIs). We reviewed the initial and final surveillance sources of 415 NIs at Hackensack Hospital during an 11-month period; cultures were taken on most of the patients with NIs. Positive microbiology cultures were the initial sources in 64.8% of NIs and secondary sources in 23.6%. Chart review of patients with positive cultures added another 2.9%. Therefore, 91.3% of NIs could have been detected by positive culture reports. The remainder were discoverable by notification from nurses, or by studying admission cards or temperature rounds. Although the NIs missed by lack of a positive culture report often were significant infections, these alternate methods of surveillance did not help to detect any clusters of infections that could be traced to a particular source among the staff or the environment.