We conducted a study to evaluate risk factors for developing typhoid fever in a setting where the disease is endemic in Karachi, Pakistan. We enrolled 100 cases with blood culture-confirmed Salmonella typhi between July and October 1994 and 200 age-matched neighbourhood controls. Cases had a median age of 5·8 years. In a conditional logistic regression model, eating ice cream (Odds ratio [OR]=2·3; 95% confidence interval [CI] 1·2–4·2, attributable risk [AR]=36%), eating food from a roadside cabin during the summer months (OR=4·6, 95% CI 1·6–13·0; AR=18%), taking antimicrobials in the 2 weeks preceding the onset of symptoms (OR=5·7, 95% CI 2·3–13·9, AR=21%), and drinking water at the work-site (OR=44·0, 95% CI 2·8–680, AR=8%) were all independently associated with typhoid fever. There was no difference in the microbiological water quality of home drinking water between cases and controls. Typhoid fever in Karachi resulted from high-dose exposures from multiple sources with individual susceptibility increased by young age and prior antimicrobial use. Improving commercial food hygiene and decreasing unnecessary antimicrobial use would be expected to decrease the burden of typhoid fever.