Seventeen state-sponsored residential math and science schools have been created across the country to direct talented teens toward STEM careers. Admission is selective, based on competitive grades, standardized test scores, and references. Most of the schools also require preadmission interviews. However, selection interviews may be challenged as being both unreliable and invalid based on rater biases and unstructured protocols. A questionnaire was returned by nine of the eleven interviewing state schools. Results indicate unfamiliarity with selection research, unaddressed interviewer biases, and failure to conduct reliability or validity studies. Reasons schools continue to interview despite such omissions and recommendations for improvement are discussed. (Contains 3 tables.)