Following reports of frequent deaths associated with jaundice and chronic liver disease among adults in a periurban community of Karachi, Pakistan, an investigation was conducted to evaluate the relationship between injections and viral hepatitis infections, to identify the reasons why patients received frequent injections, and to observe the injection practices employed in clinics. Two hundred and three adult patients were interviewed as they left each of the 18 area clinics. Practitioners were interviewed and three consecutive injections were observed at each clinic. Eighty-one per cent of patients received an injection on the day of the interview. Of the 135 patients who provided a serum sample, 59 (44%) had antibodies against hepatitis C virus and 26 (19%) had antibodies against hepatitis B virus. Patients who received more injections were more likely to be infected with hepatitis C. If oral and injected medications were equally effective, 44% of patients preferred injected medication. None of the practitioners knew that hepatitis C could be transmitted by injections. Non-sterile syringes and needles that had been used earlier in the day on other patients were used for 94% of the observed injections. Interventions to limit injections to those which are safe and clinically indicated are needed to prevent injection-associated infections in Pakistan and other low-income countries.