BACKGROUND: The pain trajectory is an early detection/prediction method for chronic postsurgical pain (CPSP). It is unclear whether a pain trajectory can predict CPSP in patients who have undergone a total knee arthroplasty (TKA). Here we investigated (1) whether CPSP can be predicted in TKA patients, and (2) the values that can be used to predict CPSP. METHODS: We studied 211 postoperative TKA patients. We calculated the pain trajectory (pain curve slope and intercept) using the patients' self-reported pain intensity values at 1, 3, 5, and 7 days post-TKA. Using structural equation modeling (SEM), we performed a multiple regression analysis to investigate appropriate prediction models for the pain trajectory. Classification and regression tree (CHAID) methodology was used to calculate values to predict CPSP by a decision tree model. CPSP (dependent variable) was defined as >30 mm on a visual analog scale for pain intensity at 1 year post-TKA. The predictor variables were pain curve slope, intercept, age, sex, body mass index, and preoperative pain intensity. RESULTS: The pain trajectory was the best fit among the models to predict pain intensity at 1 year post-TKA. When the pain curve slope (pain trajectory) was greater than 2.8, the probability of CPSP at 1 year post-TKA was 33.3%. CONCLUSION: Our results suggest that the pain trajectory could be applied to post-TKA patients and used to calculate clinical values to predict CPSP. Our findings also indicate the possibility that patients with a positive pain curve slope in the first postoperative week may need early intervention to avoid CPSP.