A retrospective hospital record-based study for Uganda was conducted in Hoima district (an area of stable malaria transmission) and in Kabale district (an area of unstable malaria transmission). In-patient and maternity ward registers for January 1997-December 1998 were reviewed and malaria admissions, mortality, abortions, stillbirths and birthweights were recorded. Data were analysed in relation to rainfall patterns. Admissions due to malaria were significantly higher at Hoima compared to Kabale hospital in 1997 (P < 0.0001). However, the situation reversed in 1998 with significantly more malaria cases registered at Kabale compared to Hoima hospital (P < 0.0001). The increase at Kabale hospital in 1998 was attributed to increased and prolonged rains during 1997 (El Niño). Pregnancy was a risk factor for admission with malaria at Kabale, but not at Hoima. Anaemia was significantly more common among patients admitted to hospital at Hoima compared to Kabale in 1997 and 1998 (P < 0.0001 and P = 0.02, respectively). The rate of low birthweight (birthweight < 2.5 kg) was significantly higher among primigravidae and multigravidae at Hoima hospital compared to Kabale hospital (P < 0.0001). There were significantly more stillbirths at Kabale compared to Hoima hospital (P < 0.0001). Routine hospital data such as birthweight and number of malaria cases can provide information on the level of malaria transmission useful for the health services to target appropriate malaria interventions and allocate resources to control outbreaks of malaria epidemics.