AIM To determine safety of usage of blood group O donors as universal donors at the Lagos University Teaching Hospital. OBJECTIVES 1. To determine the prevalence of anti-A and anti-B haemolysins among blood group O donors at the Lagos University Teaching Hospital. 2. To determine the risk of ABO Haemolytic disease of the newborn (ABO HDN) in offspring of blood group O women. STUDY DESIGN This was a cross sectional study conducted on healthy adult volunteers who attended the blood donor clinic of the Lagos University Teaching Hospital, Lagos, Nigeria. MATERIALS AND METHODS Three hundred and fifty (350) blood group O donors were screened for anti-A and anti B haemolysins using the haemolytic properties of IgG anti-A/anti-B. 1 Those sera in which there was haemolysis after incubation of known red cells of groups A and B were further titrated and the results read both visually and spectrophotometrically. Mathematical methods were adopted to derive the population prevalence of ABO genes from phenotype data. The mathematically derived prevalence obtained were then used to estimate the risk of ABO HDN in non-group O offspring of blood group O women in LUTH. RESULTS The prevalence of both anti-A and anti-B haemolysins was 30.29% (106 out of 350 donor sera). Anti-A haemolysins had a higher prevalence of 25.14% (88 out of 350) than anti-B haemolysins that had a prevalence of 14.86% (52 out of 350). The difference between prevalence of anti-A (25.43%) and anti-B (14.86%) is highly significant (p = 0.0009). Further titration of positive sera showed that anti-B haemolysins had a higher titre both visually and spectrophotometrically than anti-A haemolysins. The arithmetic mean for visual titres for anti-B haemolysins was 7.25 ± 4.08 compared with 6.27 ± 4.74 for anti A haemolysins (p = 0.2). The arithmetic mean for the spectrophotometric titres for anti B haemolysins (101.22 ± 64.45) was also higher than that for anti-A haemolysins (80.85 ± 64.83). Again the difference did not reach significant levels (p= 0.07). No relationship was found between age, sex and rhesus D blood group and the prevalence of haemolysins. Mathematical estimates from ABO phenotypic data showed that prevalence of A gene, B gene and O gene in Lagos are 0.1416, 0.1209 and 0.7375 respectively. From these estimates, the likelihood of a scenario whereby a blood group O female will be married to a non-blood group O husband to produce a non-blood group O offspring was 2 calculated to be 14.3%. Using a prevalence of ABO haemolysins of 30.29% as obtained in this study, it was calculated that 30.29% of every 14.3 deliveries will be at risk of ABO HDN. Thus approximately 4.3 out of every 100 babies delivered in LUTH are likely to suffer ABO HDN. CONCLUSION 1. The use of group O blood for non-group O recipients is to be discouraged considering the high prevalence rates obtained in this study. 2. The estimate of 4.3% of deliveries at risk of ABO HDN has to be clinically evaluated.