Background Painful knee after arthroplasty concerns up to 21% of patients at six months. We aimed to evaluate: the application of a decision tree to explain painful knee after total knee arthroplasty (TKA), the rate of unexplained pain after complete algorithmic screening. The aim of the study was to evaluate the causes of painful TKA. Our hypothesis was that it is possible to find the cause of the pain in more than 90% of cases. Methods A single-center retrospective study analyzed all 1130 consultations between 1 April 2017 and 31 July 2018. We included all patients consulting for unexplained chronic painful knee arthroplasty. Results We included 112 knees in 105 patients as unexplained painful knee arthroplasty. Final diagnostic status was no diagnosis in seven (6.3%) cases; infection in five (4.5%); instability without real dislocation in three (2.7%); placement error in two (1.8%), due to rotational problems; loosening in 25 (22.3%): 24 (21.4%) tibial and two (1.8%) femoral; polyethylene wear in nine (8.0%); periarticular pain in 37 (33.9%): 20 with quadriceps deficiency, four with iliotibial tendinitis, four with pes bursitis, six with stiffness, and three with prepatellar bursitis); zero material problems; projected pain in 21 (18.8%); and complex regional pain syndrome in three (2.7%) cases, improved by gentle physiotherapy. Conclusions The present study was original and presents the problem in the context of everyday practice, from the clinician's point of view, with an easy-to-use decision tree that can be implemented to assess painful knee in consultation. Previous article in issue