Cross-sequential methods of analysis, designed to separate age and cohort effects, were applied to data from the Swedish Adoption/Twin Study of Aging. Thirteen cognitive variables were collected at 3 times of measurement separated by 3-year intervals. Data were available from 85 individuals from monozygotic (MZ) pairs reared apart, 132 from MZ pairs reared together, 207 from dizygotic (DZ) pairs reared apart, and 178 from DZ pairs reared together (age range at first assessment: 41-84 years). Time x Cohort interactions were found for mean performance on 8 of the measures, revealing stable mean performance in the younger cohorts and longitudinal decreases in mean performance in the older cohorts. Cohort and time effects for total variance were mixed; little evidence was found for increases in variance with age. Age changes and cohort differences in genetic and environmental components of variance were test-specific; several Cohort x Time interactions attained significance. Heritability of the general cognitive ability factor showed significant longitudinal decreases over time in the older cohorts.