The ability to detect changes in patient-perceived pain after total joint arthroplasty (TJA) is critical to manage postoperative pain. The minimal clinically important difference (MCID) for visual analog scale for pain (VAS-P) has not been investigated in this population. This study investigated the MCID for VAS-P in the TJA population.Postoperative pain scores were collected on 139 total hip arthroplasty (THA) and 165 total knee arthroplasty (TKA) patients. VAS-P was measured and Likert scores for changes in pain recorded together throughout the hospitalization per patient. Using a linear mixed model, the mean difference between preceding and current VAS-P was calculated and correlated with Likert score, when the patient reported at least slight improvement or worsening in pain, defining the MCID. Minimal detectable change was calculated using the VAS-P standard error of the means for patients reporting “no change.”For THA, the overall mean and average highest VAS-P were 35.0 mm and 50.4 mm, respectively. For TKA, the overall mean and average highest VAS-P were 42.6 mm and 61.1 mm, respectively. The minimal detectable change in VAS-P was 14.9 mm for THA and 16.1 mm for TKA. The MCID for THA and TKA pain improvement was −18.6 mm and −22.6 mm, respectively, and for worsening was 23.6 mm and 29.1 mm, respectively.In the postoperative TJA population, VAS-P MCID changes depend on the type of surgical intervention, and whether pain is improving or worsening. Statistically significant VAS-P, improving −18.6 mm and −22.6 mm for THA and TKA patients, respectively, sets a reasonable threshold to identify clinically meaningful pain intervention with high specificity.