Fitting multilevel models to diary data allows the analysis of development trends, group differences, and breakpoints in time trends. The aim of this study is the application of multilevel modelling to determine time trends in the diary data of a group of obese patients with and without binge eating disorder (BED). Throughout the course of a four-month multi-modal intervention program, the patients answered questions daily on handheld computers about their eating behaviour as well as their psychological and physical states. Multilevel analysis of the diary data shows that initially the trend of the eating behaviour over time decreases but increases towards the end of the therapy. Results of the application of a relatively new method to identify break points in linear trends indicate that the 85 (th) day of the therapy is a break point in the trend over time of the eating behaviour. Significant differences in the development of the eating behaviour over time were found for the subgroups of patients with and without BED.