The jury selection computer model is designed to determine optimal jury selection performances of the variety addressed by mathematical modelers but without the limitations imposed by the assumptions on which those models are based. The optimal model is a normative model based on empirical data rather than on assumed distributions of juror characteristics and can operate on simulated jurors possessing characteristics identical to and distributed in the same way as the characteristics possessed by real jurors. These characteristics can include not only demographic, attitudinal, and personality information but also jurors' actual responses to trials. For a computer simulation of jury selection, 269 actual jurors provided verdict preferences together with attitudinal and demographic characteristics. The computer simulation applied to these data employed three types of jury selection strategies: a random strategy in which jurors can be seated without challenges or using arbitrary criteria, an optimal strategy that uses the distribution of jurors' prior probabilities of voting for conviction in combination with recursive algorithms to determine an optimal strategy, and a characteristics strategy that models the strategies employed by real attorneys. Results indicate that under the right circumstances recursive models designed to optimize jury selection can be applied in filed settings and could have a substantial impact on the composition of juries and the jury decisionmaking process. However, data used in the simulations described herein are unlike the data available to practicing attorneys. In addition, the generalizability of the results is still being tested. Finally, while the optimal model could serve as an aid to attorneys' intuitions about biases, the model is not sufficiently portable for courtroom use. A few diagrams, 3 tables, and 47 references are provided. (Author abstract modified)