A placebo-controlled crossover study of behavioral effects of triiodothyronine (T3) was conducted in 30 young clinically euthyroid autistic children. Multiple independent raters and multiple rating scales were used. Except for a few symptoms that were reduced on T3, the drug did not differ from placebo. Time itself accounted for most of the improvement in the whole sample. As a group, the lower IQ children responded to T3. The individual children who were responders could not be defined by any parameter.