Eighty-nine consecutive primary total knee arthroplasties performed for osteoarthrosis were evaluated to determine the effect of patellar thickness, alignment, joint line, patellar height, type of patellar prosthesis, and length of follow-up on patellar tilt. Preoperatively, 62% of the patellae were tilted neutrally and 38% tilted laterally on standard 45 degrees Merchant-view radiographs. At 3-week follow-up, 52% were neutral, 32% tilted laterally and 16% tilted medially. At the most recent follow-up (mean 36 months), 41% were neutral, 46% tilted laterally and 13% tilted medially. Five (5.6%) required reoperation for patellar complications. Change in joint line, in alignment, and in patellar height had no effect on postoperative patellar tilt. Patients who required a lateral release had postoperative patellar tilt measurements that were not statistically different from those who did not require a lateral release. Metal-back domed patellae were more likely to tilt laterally than all-polyethylene domed patellae; however, the metal-back patellae had a longer mean follow-up (41 months) compared with the all-polyethylene patellae (27 months). The positive correlation between preoperative patellar thickness and postoperative patellar tilt was statistically significant. The negative correlation between change in patellar thickness and postoperative patellar tilt also was statistically significant. Finally, postoperative medically tilted patellae moved toward neutral with time and laterally tilted patellae not only remained laterally tilted but also the tilt increased with time.