The maximum likelihood estimate of the shape parameter, k, of the gamma distribution is known to be biased. This study investigates the extent and nature of this bias and compare it with the crude estimate of k. Daily rainfall data are simulated, and a bias- correction procedure is applied to these data. The results of the study suggest that correction for bias is important, particularly when the number of parameters is large; nevertheless, the resulting estimates are still biased. Moreover, a crude estimate of k based on the mean deviance is the least biased over the range of values of k investigated here.