In May 1992, 81 bacteriologically confirmed cases of typhoid fever (TF) were identified in all districts of Tabuk City in northwestern Saudi Arabia. Attack rates (AR) in residential districts ranged from 0·9–10·3 per 10000. Confirmed cases included 9 workers in the city's referral hospital, King Khalid Hospital (AR 140/10000), 2 in families of medical staff, 57 in the community (AR 4·4/10000) and 13 in a local military cantonment (AR 0·8/10000). The outbreak began with the onset of TF in the three areas within 5 days, continued for 7 weeks, and ended 2 weeks after chlorination began. Among water sources, the odds ratio (OR) was highest (2·6; 95% confidence interval [CI] 1·25–5·39) for water purchased from reverse osmosis (RO) plants, especially RO plants supplied by one well (ASUW) (OR = 7·05; 95% CI 2.51–20·7). The aquifer for ASUW lay partially beneath a depression where city sewage collected. Unchlorinated water samples from ASUW 1 month after the outbreak ended yielded coliforms. ASUW probably became contaminated with Salmonella typhi when KKH demand overtaxed the aquifer and drew in surface water. Membranes in RO plants using this unchlorinated well water could then become fouled with S. typhi. RO plants, which are common throughout Saudi Arabia, need close monitoring. WTater for RO must be prechlorinated to prevent microbiologic fouling of the membranes.