Results of 170 total knee arthroplasties(TKAs) performed from 1986 to 1995, followed up for at least one year, were reviewed. Lateral retinacular release was performed in 76 knees(45%) and not performed in 94 knees(55%). We compared the results of TKAs combined with lateral retinacular release with those without lateral retinacular release. The clinical results of TKAs were evaluated by the Knee Scoring System of Hospital for Special Surgery(HSS). The average follow-up period was 34 months. The clinical results assessed by HSS score, pain score and range of motion improved postoperatively in both groups, but there was no significant difference between the two groups. There were 8(4.7%) with patellar subluxations in 170 TKAs, one(1.3%) in the group with lateral retinacular release and 7(7.5%) in the group without lateral retinacular release(t-test; p<0.05). Analysis of lateral tilt of patella on roentgenographs also showed a significant difference between the two groups. 7 knees(9.2%) with lateral tilt were observed in the group with lateral retinacular release and 22 knees(23.4%) with lateral tilt in the group without lateral retinacular release(t-test; p<0.05). There was no osteonecrosis or fracture of patella in either group. In summary, lateral retinacular release might improve the patellar tracking and lower the incidence of patellar subluxation and lateral tilt of patella. Lateral retinacular release alone did not seem to cause patellar osteonecrosis and subsequent patellar fracture which were detectable in plain roentgenographs. So, we suggest that lateral retinacular release can be performed without great risk of potential patellar complications in TKAs requiring lateral retinacular release for proper patellar tracking.