Six different hysteresis models for the moisture retention characteristic (MRC) are described and their performance evaluated using laboratory measurements and literature data. The choice of these models is based on reported accuracy in literature and their limited data requirement for calibration. Several wetting and drying curves were measured for two sand samples in three replicates. In addition, eight data sets from literature were included in the analysis. Three statistical criteria and three statistical hypotheses were used to evaluate the models' performance. This study confirms that neglecting hysteresis may result in significant errors when describing wetting and drying data. Models using one branch of the MRC for calibration purposes have an inferior performance relative to those employing two branches. The best models were the conceptual models needing two branches for calibration. Using more than the main loop for calibration did not significantly increase the accuracy of the best performing models in describing the data. It was found that only uneven order wetting and even order drying scanning curves need to be considered.