BACKGROUND AND AIMS: Retrospective measures of alcohol intake are becoming increasingly popular; however, the reliability of such measures remains uncertain. This study assessed the reliability of a retrospective decade-based life-course alcohol consumption questionnaire, based on the standardized Alcohol Use Disorder Identification Test-Consumption (AUDIT-C) administered in older age in a well-characterized cohort study. DESIGN, SETTING, PARTICIPANTS AND MEASUREMENTS: A retrospective alcohol life-grid was administered to 5980 participants (72% male, mean age 70 years) in the Whitehall II study covering frequency of drinking, number of drinks in a typical drinking day and frequency of consuming six or more drinks in a single drinking occasion in the teens (16-19 years) through to the 80s. A subsample of 385 individuals completed a repeat survey to determine test-retest reliability. Retrospective measures were also compared with prospectively ascertained information and used to predict objectively measured systolic blood pressure to test their predictive validity. FINDINGS: Across all decades of life, test-retest reliability was generally good (κ range = 0.62-0.78 for frequency, 0.55-0.62 for usual number of drinks and 0.57-0.65 for frequency of consuming six or more drinks in a single occasion). The concordance between prospective and retrospective measures was consistently moderate to high. The life-grid method performed better than a single question in identifying life-time abstainers. Retrospective measures were also related to systolic blood pressure in the manner anticipated. CONCLUSION: A retrospective decade-based AUDIT-C grid administered in older age provides a relatively reliable measure of alcohol consumption across the life-course.