The outcome in three groups of patients with bacteriologically confirmed typhoid fever caused by Salmonella typhi, treated during three episodes between 1948 and 1990 in Java, Indonesia, was compared by retrospective analysis of hospital records. The study population consisted of three groups of patients. Group I (n= 50) was treated in Batavia (the present Jakarta) from 1948 to 1950, Group II (n= 61) in Yogyakarta from 1952 to 1956, Group III (n= 105) in Semarang from 1989 to 1990. Main outcome measures were days until defervescence, early relapses during hospitalization, duration of hospital stay, complications and mortality. Group I received supportive treatment only, Group II low doses of chloramphenicol (total 12.5 g) and Group III full doses of chloramphenicol (total 27 g); occasionally other antibiotics were used. In Group I, II and III the mean number of days until defervescence was 16, 8 and 6 and the mean number of days in hospital 43, 47 and 15, respectively. Mortality was 26%, 10% and 5% and complications occurred in 38%, 18% and 13%, respectively. Between Group I and Group II the differences in mortality and complications were statistically significant (P < 0.05). Compared to Group I the proportion of early relapses was higher in Group II, but was zero in Group III. There were significantly fewer gastrointestinal complications in Group II than in Group I (P < 0.01) and even fewer in Group III. When no antibiotic against S. typhi was available, typhoid fever had a protracted course, and only 74% of patients survived. Even with low dosages of chloramphenicol, defervescence was earlier and mortality and complications decreased dramatically, but early relapses were frequent. Full doses of chloramphenicol for a sufficient period of time only slightly reduced mortality and complications further, but eliminated early relapses completely.