This study used a longitudinal design to examine age-related changes in a well-defined sample of Swedish people ranging from 86 to 92 years of age at baseline. The longitudinal design encompassed three measurement occasions with 1 year intermeasurement intervals. The results were analyzed by multivariate analyses of variance (MANOVA), which is useful for comparing individuals over time. Healthy middle-aged subjects (39 years SD +/- 5.8) served as controls. The proliferative responses to Concanavalin A (Con A), a T-cell mitogen, indicated significant lower levels in responses of the old when the two groups were compared. The MANOVA revealed no significant change in mitogen responses over measurement occasions in the old sample as compared with the young. However, when cell types and lymphocyte subpopulations were examined, significant differences were found between the two age groups in many of these parameters and for some (lymphocyte percentages and numbers, CD3 numbers) the MANOVA indicated significant decreases over the measurement occasions in the very old. The results also consistently indicated significant intraindividual correlations in cell types, lymphocyte subpopulations, and mitogen responses over time.