OBJECTIVES: Analysis of functional data from longitudinal studies is usually complicated by missing data. The use of repeated measures ANOVA for statistical analysis tends to decrease the sample size and thus affect power and significance of the results. The present study aims to address this problem with use of hierarchical regression or multilevel modeling. METHODS: We analysed functional results of 4777 patients following hip resurfacing arthroplasty. These patients were followed annually using Harris Hip score and Merle d'Aubigné score. Individual domains of Pain, mobility and range of movement were recorded. The scores comprised a follow up period of nine years with some missing values. The data was analysed using multilevel techniques in statistical package SYSTAT 11.0. Model was fitted at two levels, with level one being the scores and level two the patients. RESULTS: In all the domains of function, pain and movement, pre operative score and gender were significantly associated with post operative hip function ( p