We performed a retrospective, matched-control analysis of the results of total knee arthroplasty with cement in twenty-two patients who had had a previous patellectomy for either a fracture of the patella (sixteen patients) or severe patellofemoral osteoarthrosis or chondromalacia patellae (six patients). Nine of the patients were men and thirteen were women. The average age at the time of the total knee arthroplasty was sixty-nine years (range, fifty-nine to seventy-four years). The average time from the patellectomy to the total knee arthroplasty was eight years (range, two to fourteen years). The patients were divided into two groups according to the type of implant that had been used. Group A (nine patients) had had insertion of a posterior stabilized prosthesis (a posterior cruciate-sacrificing implant) and Group B (thirteen patients) had had insertion of a posterior cruciate-sparing implant. Two computer-generated matched groups of patients who had had a total knee arthroplasty with insertion of implants that were identical to those in the study groups but who had not had a previous patellectomy served as controls. Group C consisted of patients who had had insertion of the same type of implant as that used in Group A, and Group D consisted of patients who had had insertion of the same type of implant as that used in Group B. All patients were evaluated before the arthroplasty and five years postoperatively with use of the rating system of the Knee Society.(ABSTRACT TRUNCATED AT 250 WORDS)