One hundred eight GUEPAR knee arthroplasties have been studied with a follow-up of one year on 41; 2 years on 22 and 3 years on 45 knees. Overall results were 17% excellent, 44% good, 10% fair and 29% poor. Excellent results were comparatively fewer in rheumatoid arthritis. There was a deterioration in the quality of results of arthroplasty with longer follow-up. The incidence of deep infection was 11%. There was a significant correlation between early wound drainage and deep infection. More than half of the infected knees have not required intervenition as yet. One knee was revised and 3 had attempted arthrodesis with one successful fusion. Two patients died with septicemia. Axle migration occurred in 8 knees and femoral stem breakage in 2 knees. Loosening was found in 27% of the knees with progressive reduction in quality of the arthroplasty. Incomplete cementing predisposed to loosening. Patellar symptoms were present in 28% of the knees. Patellar subluxation and dislocation occurred in 49% of the knees. More than half of these were symptomatic. With normal patellofemoral alignment, pain was more common in the osteoarthritic knee. Use of a patellar implant with GUEPAR knee prostesis should be restricted to severely disabled patients with major fixed deformities. Mechanical failure can be minimized by proper positioning of the implant, correct alignment of the extensor mechanism and adequate cement around the entire stem.