BACKGROUND: The current American Association of Hip and Knee Surgeons (AAHKS) guidelines recommend preoperative weight loss before total knee arthroplasty (TKA) in patients with body mass index (BMI) ≥40 kg/m2. However, there is a paucity of evidence on TKA outcomes after preoperative weight loss. This study therefore evaluated predictors of preoperative and postoperative BMI changes and their impact on outcomes after TKA. METHODS: This is a retrospective review of 3058 primary TKAs at an academic institution from 2015 to 2019. BMI was collected on the day of surgery. Preoperative and postoperative BMI at 6 months and 1 year were also obtained. BMI change of ≥5% was considered clinically significant. Mean follow-up was 3.2 years. Patient demographics, acute postoperative outcomes, and all-cause revisions were compared between patients who gained, lost, or maintained weight using univariate and multivariable analyses. RESULTS: Preoperative weight loss was predictive of postoperative weight gain (P < .001), and preoperative weight gain was predictive of postoperative weight loss (P < .001). Cox regression analysis revealed that ≥5% BMI loss preoperatively increased risk for all-cause revisions (P = .030), while ≥5% BMI gain postoperatively increased risk for prosthetic joint infections (P = .016). Patients who lost significant weight both before and after surgery had the highest risk for all-cause revisions (P = .022). CONCLUSION: Weight gain postoperatively was associated with inferior outcomes. Significant weight loss before surgery led to a "rebound" in weight gain, and independently increased risk for all-cause revision. Therefore, current recommendations for weight loss before TKA in morbidly obese patients should be re-evaluated.