Transient rises in parasitaemia occur commonly during the treatment of falciparum malaria but their prognostic significance has not been well defined. Twelve-hourly parasite counts from 133 malaria patients who were ultimately cured were therefore compared with counts from 97 therapeutic failures to determine if increase in parasitaemia was a useful early indicator of poor treatment response. Parasitaemia in both groups frequently rose during the initial 12 h of therapy (41% of all patients), but rising counts thereafter were rarer in treatment successes (P < 0.01). The relative risk of treatment failure was 3.8 if the count was higher at 24 h than 12 h, rose to 7.8 for increases at 48-60 h, and was 13.7 and 19.4 for counts above admission levels at 48 h and 60 h. These data suggest a way to identify patients at high risk of treatment failure.