A study was conducted in Kassena-Nankana district, Ghana from April 1994 to April 1995 to assess the impact of insecticide-treated bednets on malaria and anaemia in pregnancy. A secondary objective was to assess the impact of insecticide-treated bednets on pregnancy outcome, although it was recognised that the sample size would be relatively small for this purpose. The study took place within a large-scale controlled trial designed to assess the impact of insecticide-treated bednets on child mortality. The study area was divided into 96 clusters of compounds and of these 48 clusters were randomly selected to receive the intervention (treated nets). The endpoints of the study were haemoglobin levels, parasitaemia and parasite density and weight of newborns which were recorded within 7 days of delivery. All pregnant women were included in the study but the target group of main interest was primigravidae and secundigravidae. A total of 2812 pregnant women, of all parities, were enrolled into the study; 1961 (70 percent) women were seen at least once at a study clinic of whom 641 were seen for a second time between 28-40 weeks gestation. At both clinic visits, blood was taken for haemoglobin determination and malaria parasitology. Chloroquine ELISA assays, using dried blood spots on filter paper, were performed for 64 percent of pregnant women at their first clinic visits. Data were obtained on 821 delivery outcomes, including 799 newborn weights recorded within 7 days of delivery. A cross-sectional survey was done to determine the distribution of haemoglobin levels and malaria parasitaemia in non-pregnant adult females. Focus group discussions were conducted to assess study women's attitudes to antenatal care and use of bednets. Bednet use provided no protection against anaemia defined as (HbclOO g/l), severe anaemia (Hb<70 g/l), P. falciparum parasitaemia (high: >2000 parasitcs/pl or low : >0 parasite/pl) or low birth weight. The characteristics of women in the treated and the no net groups were comparable. Net usage was lower than expected, especially in primigravidae. Effective net use was as follows: primigravidae 42 percent (net coverage: 60 percent, use: 70 percent), secundigravidae 58 percent (net coverage: 80 percent, use: 72 percent), multigravidae 63 percent (net coverage: 86 percent, use: 73 percent). Below 10 percent of chloroquine ELISA assays were positive with no differences by treatment arms, parity or season. Odds ratios (ORs) for the different endpoints for those with nets in comparison to those without nets, based on an intention-to-treat (ITT) analysis did not show a statistically significant protective effect; First clinic visit: Anaemia Severe anaemia Low Parasitaemia High Parasitaemia -0.97 (0.86, 1.10) -0.91 (0.57, 1.43) - 1.13(0.54,2.38) -0.98 (0.85, 1.12) The following results were obtained when data for first and second clinic visits were combined (combined data) and restricted to one record per woman with at least 26 weeks gestation and including all second clinic visit records. Combined data: Anaemia - 0.88 (0.70, 1.09) Severe anaemia - 0.80 (0.55, 1.16) Low Parasitaemia - 0.89 (0.73, 1.08) High Parasitaemia - 1.11 (0.93, 1.33) Low birth weight: Adjusted (<2500g)- 0.87 (0.63, 1.19) Unadjusted - 0.88 (0.61, 1.24) Analysis of protection at individual level showed similar results. Based on the findings of this study, insecticide-treated bednets are not recommended as a primary tool for malaria control in pregnancy in northern Ghana. Further operational research is required to assess the impact of insecticide-treated bednets combined with chemoprophylaxis and behavioural interventions on malaria in pregnancy.