The Oxford hip and knee scores are used to measure the outcome after primary total hip and knee replacement. We propose a new layout for the instrument in which patients are always asked about both limbs. In addition, we have defined an alternative scoring method which accounts for missing data. Over a period of 4.5 years, 4086 (1423 patients) and 5708 (1458 patients) questionnaires were completed for hips and knees, respectively. The hip score had a pre-operative median of 70.8 (interquartile range (IQR) 58.3 to 81.2) decreasing to 20.8 (IQR 10.4 to 35.4) after one year. The knee score had a pre-operative median of 68.8 (IQR 56.2 to 79.2) decreasing to 29.2 (IQR 14.6 to 45.8). There was no further significant change in either score after one year. As a result of the data analysis, we suggest that the score percentiles can be used as a standard for auditing patients before and after operation.