The median absolute deviation is widely used as a robust measure of scale. In this paper, computer simulations were used to calculate finite-sample correction factors for sample sizes to n = 100. Prediction equations for these correction factors are also provided and are shown to eliminate systematic biases between even and odd values of n.