If patients could recall their physical status before total hip (THA) or knee arthroplasty (TKA) accurately it could have valuable applications both clinically and for research. This study evaluated the accuracy of a patient’s recollection one year after either THA or TKA using the Oxford hip or knee scores (OHS and OKS). In total, 113 patients (59 THA, 54 TKA) who had completed the appropriate score pre-operatively were asked to complete the score again at a mean of 12.4 months (standard deviation (. sd. ) 0.8) after surgery, recalling their pre-operative state. While there were no significant differences between the actual and recalled pre-operative scores (OHS mean difference 0.8, . sd. 6.21, 95% confidence interval (CI) -0.82 to 2.42, p = 0.329; OKS mean difference -0.11, . sd. 7.34, 95% CI -2.11 to 1.89, p = 0.912), absolute differences were relatively large (OHS, 5.24; OKS, 5.41), correlation was weak (OHS r = 0.7, OKS r = 0.61) and agreement between actual and recalled responses for individual questions was poor in half of the OHS and two thirds of the OKS. A patient’s recollection of pre-operative pain and function is inaccurate one year after THA or TKA. Cite this article: Bone Joint J 2015;97-B:1070–5