Objective To compare the treatment effect of patellar resurfacing versus patellar non-resurfacing in total knee arthroplasty. Methods We identified eligible studies in PubMed (1950 to 2008.6), OVID MEDLINE (1950 to 2008.6), OVID CINAHL (1950 to 2008.6), OVID EBM (2nd Quarter 2008), CBMdisk (1978 to 2008.6), and CNKI (1981 to 2008.6), and handsearched some Chinese orthopedic journals to identified randomize controlled trials (RCTs) comparing patellar resurfacing versus patellar non-resurfacing in total knee arthroplasty. Data were extracted and methodological quality was critically assessed by two reviewers independently. Meta-analyses were performed using Stata 10.0 software. Outcomes of interest included the number of reoperations for patellofemoral problems, the prevalence of postoperative anterior knee pain, and the improvement in HSS knee score. Results Thirteen RCTs involving 1 566 patients were included. The scores of methodological quality respectively were more than 13. The results of meta-analyses showed that patellar resurfacing could decrease the re-operation rate for patellofemoral problems (RR=0.30, 95%CI 0.14 to 0.62, Plt;0.01). The rate of postoperative anterior knee pain and the improvement in HSS knee score were comparable between patellar resurfacing and patellar non-resurfacing. Conclusion The outcome identified is re-operations for patellofemoral problems. The resurfaced patella performs better, and we find an increased relative risk for re-operation when the patella is left un-resurfaced. No differences are observed between the two groups for the prevalence of postoperative anterior knee pain, and the improvement in HSS knee score. Further well-designed and large-scale RCTs are required to determine the effects of patellar resurfacing and non-resurfacing on these outcomes.