The use of complementary or alternative treatment approaches in children with autism spectrum disorders (ASDs) is increasing, and the most popular of such approaches are diets that eliminate either gluten or casein, or both. The popularity of these diets indicates a need for more rigorous research into their efficacy. Owing to significant methodological flaws, the currently available data are inadequate to guide treatment recommendations. The purpose of this review is to examine the available trials of gluten/casein diets in children with ASDs regarding the strength of their findings and also concerning points that may be useful in the design of future studies. Seven trials of these diets in ASD are critically reviewed; 6 of these were uncontrolled trials and 1 used a single-blind design. All reported efficacy in reducing some autism symptoms, and 2 groups of investigators also reported improvement in nonverbal cognition. Design flaws in all of the studies weaken the confidence that can be placed in their findings. Careful double-blind, placebo-controlled studies are needed to evaluate whether actual benefit undergirds the diets' popularity and to provide better guidance to clinicians and caregivers. The literature currently available suggests that diets eliminating both gluten and casein (rather than either alone) should be studied first and that outcome measures should include assessments of nonverbal cognition.