A meta-analysis (MA) based on original patient data has been performed comparing low molecular weight heparins (LMWH) with unfractionated heparin (UFH) in thrombosis prophylaxis after major surgical interventions. The analyses have been done for the following prespecified groups of studies: all studies, studies in orthopaedic surgery (OS) and studies in general surgery (GS, with further separation into low-dose studies [GS-LD] and high-dose studies [GS-HD]). Deep vein thrombosis (DVT, all locations) and wound haematoma were used as primary endpoints for efficacy and safety, respectively. The analysis confirms the results of previous publication-based meta-analyses. In GS there is no relevant difference between LMWH and UFH regarding efficacy; the safety results strongly depend on the dosage: under low-dose LMWH the risk of wound haematoma is significantly lower, under high-dose LMWH it is significantly higher than under UFH. However, most of the studies in the last group used regimens of LMWH that are not considered appropriate any more. In OS there is a trend towards a better efficacy and safety of LMWH. In addition, LMWHs are superior to UFH, in OS, with respect to the secondary endpoints proximal DVT and pulmonary embolism. The rates of proximal DVT and pulmonary embolism, respectively, are consistently lower under LMWH than under UFH, whereas slightly smaller rates of distal DVT are observed under UFH.