OBJECTIVE Diagnosis and treatment of haemolytic disease of the fetus has considerably progressed since the introduction of Liley's diagram. Amniocentesis and cordocentesis have changed diagnostic and therapeutic options. Recently, some authors pleaded for restraint in diagnostic cordocentesis. In this context of relative controversy, we wanted to compare our results with those of the literature. SUBJECTS AND METHODS Thirty-nine pregnancies complicated by antigen incompatibilities were referred to our unit. The haemolytic disease was evaluated by the measurement of antibody titers, by spectrophotometry in the amniotic fluid, by measurement of fetal haematocrit in cord blood samples and by ultrasound examination. Sixty-four amniocenteses, 85 cordocenteses and 25 in utero transfusions were performed. RESULTS Alloimmunization anti-D represented 67% of the cases in our series of 39 pregnancies. Fifteen percent of the fetuses were antigen negative. One neonatal death after chorioamnionitis was observed after cordocentesis, the fetal loss rate related to the procedure was 1.2%, Six fetuses had a haematocrit below 30% at the first sampling; 9 other fetuses developed an anaemia later in pregnancy. Six fetuses underwent in utero transfusion. One of these fetuses had hydrops at the ultrasound before the procedure. Twenty-five in utero transfusions were uncomplicated in spite of the observation of one post-transfusional haematoma of the umbilical cord. The delta OD450 measurement did not predict the severity of fetal anaemia in all cases. CONCLUSION In our experience, the fetal haematocrit measurement remains the most reliable method to evaluate the severity of the haemolytic disease.