OBJECTIVE: To estimate the effect of the severity of maternal anaemia on various perinatal outcomes. DESIGN: A cross-sectional study. SETTING: Labour Ward, Muhimbili National Hospital, Dar es Salaam, Tanzania. METHODS: The haemoglobin of eligible pregnant women admitted for delivery between 15 November 2002 and 15 February 2003 was measured. Data on socio-demographic characteristics, iron supplementation, malaria prophylaxis, blood transfusion during current pregnancy, and current and previous pregnancy outcomes were collected and analysed. Anaemia was classified according to the World Health Organization (WHO) standards: normal--Hb > or = 11.0 g/dl; mild--Hb 9.0-10.9 g/dl; moderate--Hb 7.0-8.9 g/dl; and severe--Hb < 7.0 g/dl. Logistic regression analysis was performed to estimate the severity of anaemia. The following outcome measures were used: preterm delivery (<37 weeks), Apgar score, stillbirth, early neonatal death, low birth weight (LBW) (<2500 g) and very low birth weight (VLBW) (<1500 g). RESULTS: A total of 1174 anaemic and 547 non-anaemic women were enrolled. Their median age was 24 years (range 14-46 years) and median parity was 2 (range 0-17). The prevalence of anaemia and severe anaemia was 68% and 5.8%, respectively. The risk of preterm delivery increased significantly with the severity of anaemia, with odds ratios of 1.4, 1.4 and 4.1 respectively for mild, moderate and severe anaemia. The corresponding risks for LBW and VLBW were 1.2 and 1.7, 3.8 and 1.5, and 1.9 and 4.2 respectively. CONCLUSION: The risks of preterm delivery and LBW increased in proportion to the severity of maternal anaemia.