Objective Use the method of Cochrane systematic review to evaluate the difference of two fixation methods of femoral component in hip arthroplasty in order to choose the most appropriate fixation method. Methods Searches was applied to the following electronic databases: Chinese Bimedical Database (CBM) (1979 to Dec. 2004), MEDLINE (1966 to Feb. 2005), EMBASE (1984 to 2004) and The Cochrane Library (Issue 4, 2004). We handsearched Chinese Journal of Orthopaedy, the Journal of Bone and Joint Injury and Orthopaedic Journal of China (from establishment to Feb. 2005). Randomized controlled trials (RCT) were indentified and we applied RevMan 4.2 for statistical analysis. Results Nine RCTs involving 1 075 hips were included. The combined results of meta-analysis showed that the embolization occurred more commonly in the first and second generation cemented group (OR 0.02, 95%CI 0 to 0.11, P＜0.000 01), but this difference was not seen between the third generation cemented group and uncemented group (OR 0.80, 95%CI 0.36 to 1.78, P=0.58); The subsidence of femoral component (OR 12.20, 95%CI 3.58 to 41.54, P＜0.000 1) and the cortical hypertrophy (OR 69.97, 95%CI 27.88 to 175.57, P＜0.000 01) were more commonly found in uncemented group; the thigh pain occurrence, the revision for the femoral component’s cause and heterotopic ossification were found no significant difference in the two groups. Conclusions Compared with noncemented group, we found that cemented fixation may be more associated with embolism in the first and second generation cemented technique and less with femoral subsidence and cortical hypertrophy. There was no significant difference in embolization between the third generation cemented technique group and noncemented group. However, more randomized controlled trials to evaluate the occurrence of the postoperative thigh pain, revision and heterotopic ossification are needed.