Three consecutive randomized open studies have been carried out to determine the optimal dosage of low molecular weight heparin (LMWH) in the prevention of postoperative thrombosis in general surgery (892 patients). All patients undergoing abdominal, gynaecological, thoracic or urological surgery were over 40 years old and presented at least one of the following risk factors for thrombosis: previous thromboembolism, obesity, varicose veins, malignancy (30 per cent), pre-operative hospitalization over 5 days, oestrogen therapy, chronic cardiac disease or bronchitis. Isotopic venous thrombosis and bleeding complications were assessed after subcutaneous administration of a LMWH fragment (LMWH, Enoxaparine) or unfractionated heparin (UH). The three studies compared 3 X 5000 units UH daily with 1 X 60 mg, 1 X 40 mg, 1 X 20 mg LMWH daily. Thromboembolic events rates were not significantly different from group to group (UH: 3.8 per cent, 2.7 per cent, 7.6 per cent respectively compared with LMWH: 2.9 per cent, 2.8 per cent, 3.8 per cent). Bleeding episodes including wound haematoma formation, perioperative blood losses and systemic haemorrhage were not significantly different in patients receiving LMWH or UH. Significant decreases in haematocrit and haemoglobin were only observed in patients receiving 60 mg Enoxaparine (as compared to UH). An analysis using the 'intention to treat' approach gave results consistent with those of an analysis of good compliers. An overview of isotopic thromboses in the three studies gave no evidence of differences amongst the effects of the three doses of LMWH (P = 0.20), and pooling the results of the three studies using the Mantel-Haenszel procedure gave no evidence of a global difference between Enoxaparine and UH (P = 0.54). These results suggest that an optimal dosage of 20 mg/day of Enoxaparine is safe and effective in the prevention of postoperative thrombosis in this population.