As part of a study of drug treatment of aggressive behavior to be reported separately, we have evaluated the double-blind procedure in a recently completed comparison of the efficacy of lithium carbonate versus placebo in modifying aggressive behavior in nonpsychotic incarcerated delinquents. We conclude that the side effects of lithium carbonate are sufficient to reveal the medication to most subjects receiving it. Thus, while the study staff could not identify lithium-receivers at better than chance levels, and while subjects who received placebo could not identify their medication at better than chance levels, subjects who received lithium could accurately identify it. On a weekly symptom check list there was no difference between lithium and placebo groups on average lithium target symptoms reported during 4-week pre- and postmedication control periods; however, lithium-receivers reported significantly more target symptoms every week medication was administered. Of 16 subjects who quit the study, 14 had received lithium and nearly all of those who gave reasons for quitting specified side effects, most often nausea. The methodological problems of using lithium in a double-blind design might be overcome by employing a “discontinuation” design, or, speculatively, a double-blind, cross-over design utilizing an “active placebo”.