The outcome at 10 years is reported of a prospective study of 2 cohorts of total knee arthroplasties treated with (center A) or without (center B) patellar replacement. The same tibiofemoral components were used in all knees. The cohorts were demographically similar. A total of 124 patellae were treated by replacement, and 143 were treated without replacement. The clinical outcome and the patellofemoral revision rates were the same in the 2 cohorts: 1 patient required analgesia for anterior knee pain after replacement, and 1 without replacement required patellar replacement for pain. In the replaced group, patellofemoral survival on a best-case scenario was 100% at 10 years; on a worst-case scenario, 96%. One of the unreplaced patellae had been resurfaced for pain by 10 years. In view of the satisfactory and similar outcomes with and without replacement, we suggest that an appropriate design for the prosthetic trochlea, rather than the replacement or otherwise of the patella, is the main determinant of patellofemoral outcome in total knee arthroplasty. Patella replacement may be optional. Desirable trochlea design features are described.