The frequency of Salmonella-infected food workers identified through routine surveillance from 1997 to 2004 in Minnesota was determined in order to evaluate the impact of surveillance on the detection of outbreaks in restaurants and to quantify the duration of Salmonella shedding in stool. Of 4,976 culture-confirmed Salmonella cases reported to the Minnesota Department of Health, 110 (2.2%) were identified as food workers; this was less than one-half the number expected based on the incidence of Salmonella in the general population. Twenty food workers (18%) were associated with outbreaks. Twelve were involved in nine independent outbreaks at the restaurants where they worked. The identification of the index food worker in six of these outbreaks was critical to the initiation of outbreak investigations that revealed much larger problems. Among food workers who submitted specimens until at least one negative result was obtained (n = 69), the median duration of shedding was 22 days (range, 1 to 359 days). Among the four most common serotypes (Enteritidis, Typhimurium, Heidelberg, and Newport) the median duration of shedding was significantly longer for Salmonella Newport (80 days; P = 0.02) and for Salmonella Enteritidis (32 days; P = 0.04) than for Salmonella Heidelberg (8 days). Food workers should be considered an important source of Salmonella transmission, and those identified through surveillance should raise a high index of suspicion of a possible outbreak at their place of work. Food service managers need to be alert to Salmonella-like illnesses among food workers to facilitate prevention and control efforts, including exclusion of infected food workers or restriction of their duties.