Three versions of a hypothetical insanity-defense case called Albert were randomly distributed to a sample of 1,002 psychiatrists and clinical psychologists, with usable returns being received from 262 subjects. A favorable opinion of Albert's insanity defense was found to correlate with being a psychiatrist, having a liberal ideology, being in favor of the insanity defense in general, and having received the 'neutral,' as opposed to the 'sympathetic' or 'unsympathetic,' version of the case. A post hoc analysis also found that coming from a state that placed the burden of proof on the prosecution was associated with a favorable opinion of Albert's insanity defense. Taken together, these variables accounted for 42 percent of the nonerror variance in opinions of Albert's insanity defense. (Publisher abstract)