An alternative item-selection procedure for use with Owen's Bayesian adaptive testing strategy is proposed. This procedure is, by design, faster than Owen's original procedure because it searches only part of the total item pool. Item selections are, however, identical for both methods. After a conceptual development of the rapid-search procedure, the supporting mathematics are presented. In a simulated comparison with three item pools, the rapid-search procedure required as little as one-tenth the computer time as Owen's technique. (Author)