Four forms of a selective reminding test were administered to 20 male and 20 female subjects 1 week apart in one of four orders determined by a Latin square. On many of the dependent measures, Form 1 was significantly more difficult than the other three forms, which were equivalent in difficulty. For many of the measures, performance on the first test administered was significantly lower than that on the third and fourth tests administered. Performance on the 2nd, 3rd, and 4th tests administered was similar. Males made significantly more intrusions than females. Interclass correlation coefficients ranged from .414 to .654. Implications for clinical use of the selective reminding tests are discussed.