Simulated tailored tests were used to investigate the relationships between characteristics of the item pool and the computer program, and the reliability and bias of the resulting ability estimates. The computer program was varied to provide for various step sizes (differences in difficulty between successive steps) and different acceptance ranges (precision with which selected items must match desired item difficulties). Item analysis and ability estimates were based on a one-parameter logistic (Rasch) model. Item pools varied in size from 9 to 181 items, and in shape to include normal, bimodal, rectangular, and postively skewed distributions of item difficulty. In addition, item difficulty distributions were designed to fit either precise theoretical distributions or actual test distributions. The acceptance range parameter was important because it influenced the point at which testing was discontinued; there was a fixed maximum of 20 items but a minimum in some circumstances of 4 items. Two different tailored test programs were used, TREE1P and SIM1P. Results were based on 25 simulated administrations under each of the various conditions. (CTM)