In this project, we plan to perform research on inference of change-points in longitudinal data. We propose to apply a mixed-effects regression model to include between- and within-subject variation as well as effects of covariates. Then likelihood methods will be used to estimate model parameters including the change-points and the EM algorithm will be used for computation. For an early detection of a change in sequentially observed data, we plan to generalize a cusum procedure to incorporate covariates and unrestricted covariance structure. When there are only a few repeated measurements taken for each subject, repeated confidence intervals will be derived.