OBJECTIVES: Surgical options for patients with symptomatic bilateral knee osteoarthritis are 1) simultaneous bilateral total knee arthroplasty (BTKA) under one anesthesia and 2) staged total knee arthroplasty (STKA) with two distinct operations separated by a few days up to one year. A number of studies have compared post-operative complications after BTKA versus STKA by simply collecting and then contrasting outcomes collected retrospectively. However, this methodology is biased because it fails to account for the patients who had STKA planned but who never completed the second stage because they died or developed a serious post-operative complication after the first operation, leading to cancellation of the second STKA. The purpose of this study was to demonstrate the misclassification bias associated with simply comparing operative outcomes after BTKA versus STKA. METHODS: : To demonstrate the bias, a mathematical derivation and graphical presentation were developed.  RESULTS:   First, we mathematically demonstrated that the observed proportion of complication (Pobserved) in patients who completed both STKA operations underestimates the true proportion of complication (Ptrue). Second, we graphically demonstrated that STKA always appears to be safer than BTKA even if when the proportion of post-operative complications observed is held constant. When data were simulated using a true odds ratio of 1, the observed odds ratio ranged from 0.899 to 0.557 for various combination of other probabilities.CONCLUSIONS: Most published studies have reported that post-operative complications are lower for STKA compared with BTKA. However, our analysis indicates that any conclusions based simply on retrospective analysis of subjects who successfully completed STKA is biased because it includes only cases that recovered after the first operation rather than all of the patients that had STKA planned. Absent a prospective study, the only fair and unbiased comparison of post-operative complications between STKA and BTKA requires adjustments to account for this bias.