To review our experience with nonoperative versus operative management of patients with patellar delayed union or nonunion. Retrospective study with an average follow-up of 64 months (range 5-135) after definitive treatment. All patients were reviewed and evaluated at a large multi-specialty clinic. The series represents twenty patients who all presented to our institution with a diagnosis of patellar nonunion irrespective of their initial treatment. This included twelve males and eight females with an average age of 38 (range 12-76) years. Initial treatment of the original fracture was nonoperative in 12 and operative in eight. All fractures progressed to symptomatic nonunion at an average of 34 months from original injury (range 4-109). Definitive treatment of the nonunion was nonoperative in seven patients and operative in 13. Nonoperative management consisted of observation, activity modification, physical therapy, and local pain relief measures. Operative management included open reduction and internal fixation, partial patellectomy, or patellectomy. The internal fixation consisted of tension band wiring, Bunnell wiring, cerclage wiring, or screw fixation. MAIN OUTCOME OR MEASURES: Patients were reviewed for radiographic analysis as well as Knee Society knee and function scores. Definitive treatment was nonoperative in seven patients. Their mean Knee Society knee and function scores at the time of presentation with nonunion were 72 and 78, respectively, with an average knee range of motion of 127 degrees. The nonunions of thirteen patients were treated operatively. Knee Society knee and function scores at the time of presentation with nonunion averaged 82 and 80, respectively, with an average knee range of motion of 112 degrees. Patients who had operative management or elective nonoperative management performed better than those who refused operative management. Patients treated surgically had an average Knee Society score of 94, a function score of 93, and an average knee range of motion of 109 degrees. Those treated nonoperatively had an average knee score of 83, a function score of 75, and an average range of motion of 120.0 degrees. In the nonoperative group, all seven patients had persistent radiographic nonunion. Only one of thirteen patients treated operatively had persistent radiographic nonunion. Our findings suggest that patients with minimally symptomatic delayed union or nonunion of the patella can be successfully treated nonoperatively with the knowledge that the fracture will not unite. Operative management of symptomatic patients can be expected to achieve union and increase function of the knee.