Technical Abstract: With greater emphasis being placed on management of potential risks of natural enemy releases for biocontrol programs and the need to justify research budgets, the efficient selection of effective natural enemies is increasingly important. Historically there has been little agreement regarding how or whether this can be accomplished. Recent studies have demonstrated that there is good correspondence between insect host-finding behavior in well-designed laboratory studies and their performance of this behavior in the field. Success in measuring efficacy of candidate agents remains somewhat of an art due to the multitude of factors influencing efficacy, but will be improved by attention to the following: 1) Natural enemy candidates must be characterized by morphological taxonomy or genetic markers at the onset of a program, 2) Attention should be devoted to candidates from matching climates when possible, and 3) Quarantine evaluations should be followed by semi-field or field cage evaluations whenever possible before proceeding with widespread releases. The application of these principles is discussed in regard to U.S. biocontrol programs for Bemisia tabaci (Gennadius) and Aphis glycines Matsumura. Proper project planning and interdisciplinary cooperation will enhance the chances for a successful project. Having progressed from a 'one man show' to a team of interdisciplinary collaborators, the art and the science of biocontrol is steadily decreasing the risk of releasing ineffective agents.