A comparison of 135 resurfacing with 116 inset biconvex patellar components with a single total knee arthroplasty design was performed. At 2.5 +/- 0.5 years postoperatively, patellar tilt in the resurfacing group was 4 degrees +/- 6 degrees compared with 0.5 degree +/- 5 degrees in the biconvex group (p < 0.0001). Lateral patellar subluxation was 6% +/- 10% in the resurfacing group compared with 3% +/- 9% in the biconvex group (p < 0.05). Lateral release was required in 79% of the resurfacing group compared with 56% in the biconvex group (p < 0.0001). Complete patellar radiolucent lines were seen adjacent to 4 biconvex and 2 resurfacing patellar prostheses. Patellar radiolucent lines were seen more frequently in area 1 in the biconvex group compared with the resurfacing group (p < 0.01). An inset patellar implant appears to provide better radiographic alignment than a resurfacing implant, but it has a higher incidence of radiolucent lines.