This article reports the results of a regression-based cohort analysis of physician utilization rates among eight 4-year cohorts of elderly Americans over an 8-year period. Data on 99, 445 noninstitutionalized individuals aged 56–95 were taken from the pooled 1972–73, 1976–77, and 1980–81 Health Interview Surveys. The parameters of the behavioral model of health services utilization were then estimated separately for each cohort within each pooled survey. The partial unstandardized regression coefficients obtained were used to construct standard cohort tables, which were then visually and statistically compared to identify any changes due to aging, period, or cohort effects. The findings suggest that the declining rate of physician utilization among the oldest-old results from an accelerated decreased response to health-related limitations in activity. Six plausible explanations for this changing response are considered.