Longitudinal studies consist of measurements of individuals repeatedly and irregularly measured on a large number of features across time, and commonly have missing values. Data with missing values may significantly reduce usable sample size. Moreover, analyzing incomplete data sets can lead to biased and misleading results. Imputation methods, such as multiple imputation, regression imputation, and machine learning imputation can be applied to impute missing values, however each method is valid under certain conditions. In this study, commonly used imputation methods specifically for longitudinal data will be reviewed, compared, and evaluated. The performance of imputation will be assessed through both simulation studies and real data applications. Results from this study will help guide researchers in selecting the most appropriate imputation method for their own data to improve the data quality, hence reduce the potential bias in research findings.