AIMS In large randomized trials, thromboprophylaxis with fondaparinux in major orthopaedic surgery (MOS) has been shown to be superior to low molecular weight heparin (LMWH) prophylaxis with comparable safety. However, patients treated under trial conditions are different from unselected patients and efficacy and safety outcomes may be different in unselected patients in daily practice. We performed a retrospective cohort study to compare the efficacy and safety of venous thromboembolism (VTE) prophylaxis with fondaparinux or LMWH in 3896 consecutive patients undergoing major orthopaedic surgery at our centre. METHODS All patients undergoing MOS between January 2006 and December 2009 were retrospectively analyzed using patient charts, hospital admission and discharge database, quality management database, transfusion unit database and VTE event documentation. VTE standard prophylaxis at our institution was LMWH (3000-6000 aXa units once daily) from January 2006 to December 2007 or fondaparinux 2.5mg from January 2008 to December 2009. In these two large cohorts of unselected consecutive patients, in-hospital incidences of VTE, surgical complications, severe bleeding and death were evaluated. RESULTS Symptomatic VTE was found in 4.1% of patients in the LMWH group (62/1495 patients; 95% CI 0.032, 0.052) compared with 5.6% of patients receiving fondaparinux (112/1994 patients, 95% CI 0.047, 0.067; P= 0.047). Distal deep vein thrombosis (DVT) was significantly more frequent in the fondaparinux group (3.9%, 95% CI 0.031, 0.048; vs. 2.5%; 95% CI 0.018, 0.034; P= 0.021). No significant differences in the rates of major VTE or death were found. Rates of severe bleeding, transfusion of RBC concentrates, plasma and platelet concentrates were comparable between both treatment groups. However, patients receiving fondaparinux had significantly lower rates of surgical revisions (1.6%, 95% CI 0.011, 0.022 vs. 3.7%, 95% CI 0.028, 0.047; P < 0.001). Multivariate analysis revealed previous VTE (HR 18.2, 95% CI 11.6, 28.5; P < 0.001) and female gender (HR 1.9, 95% CI 1.3, 2.7; P < 0.001), but not fondaparinux prophylaxis (HR1.3, 95% CI 0.9, 1.7; P= 0.184) to be associated with significantly increased VTE risk. DISCUSSION Thromboprophylaxis with fondaparinux is less effective to prevent distal VTE than LMWH in unselected patients undergoing MOS, but is equally effective with regard to rates of major VTE and death. However, differences in efficacy of LMWH or fondaparinux are of little relevance compared with a history of VTE or female gender, which were found to be the main VTE risk factors in MOS. The safety profile of fondaparinux was comparable with LMWH with regard to rates of severe bleeding complications, but patients receiving fondaparinux had significantly less surgical complications requiring surgical revisions. Both our efficacy and safety findings differ from data derived from large phase III trials testing fondaparinux against LMWH in MOS, where overall rates of symptomatic VTE were lower and the safety profile of fondaparinux was different. CONCLUSION We conclude that the strict patient selection and surveillance in phase-III trials results in lower VTE and bleeding event rates compared with unselected routine patients. Consequently, the efficacy and safety profile of thromboprophylaxis regimens needs to be confirmed in large registries or phase IV trials of unselected patients.