A least squares procedure was used in fitting, in which the standard error for the difference between the moisture contents from the experimental data and each of the drying models was minimized. The fitted drying constants were found to be 0.11 h_1 and 0.09 h_1 for continuous and intermittent drying, respectively. The fitted activation energy and the estimated diffusion coefficient were 13.4 kJ/mol and 1.09 ท 10_9 m2/s, respectively. The diffusion model was found to fit the moisture contents better than the empirical model. This was shown by the standard errors obtained, which were 50% and 88% lower than those obtained in the fitting of the empirical model for continuous and intermittent drying, respectively. The analysis of the drying rates using the concept of a characteristic drying curve showed that this concept only gave a reasonable fitting of moisture contents for continuous drying. It was also found that the sugar content of the bananas at different degrees of ripeness had an insignificant effect on the characteristic drying curve. Overall, the results of this work suggest that the empirical model and the characteristic drying curve are applicable for describing the drying kinetics of banana dried continuously but not intermittently. On the other hand, the diffusion model, which includes the variation of moisture content and temperature throughout the banana in its solution, describes the drying kinetics of banana well for both continuous and intermittent drying. This makes the diffusion model suitable for predicting the relaxation processes that occur when the drying conditions are interrupted, such as in intermittent drying.