Prophylactic use of antimicrobics represents about 30% of total hospital use of these agents. Previous studies in animal models have suggested that timing of such prophylaxis is crucial, to the extent of the existence of a "critical period" for effective administration. This phenomenon has further been shown to exist in man, in numerous double-blind, controlled clinical trials in various types of surgical wounds. On these bases, we introduced new prophylactic protocols in a university hospital, specifically directed at reducing infection in high-risk procedures. Surveillance was conducted by infection control nurses, and one antibiotic-day was the comparative unit. Infection rates were markedly diminished under these new protocols. Total usage of antimicrobial drugs also decreased by 38% over previous rates. The rationale for limiting prophylaxis to a circumscribed period was clearly demonstrated, although the exact mechanisms responsible for this effect are, as yet, unclear.