OBJECTIVE: The performance of the Alcohol Use Disorders Identification Test (AUDIT) in screening for heavy drinking among the elderly has been unsatisfactory. The aim of the present study was to determine whether tailoring the cut point improves the performance of the AUDIT and its derivatives in this age group. METHODS: From a stratified random sample of 804 Finns aged 65-74 years, 517 subjects (64.3%) completed the AUDIT and the Timeline Follow-back (TLFB) interview regarding alcohol consumption. A subject was defined as a heavy drinker if consumption of ≥8 drinks (approx. 12 g) on average in a week or ≥4 drinks at least in 1 day during the prior 28 days was reported. Combinations in which both sensitivity and specificity are ≥0.8 were defined as optimal. The elderly specific AUDIT-3 is a modification in which the binge drinking threshold is ≥4 drinks. RESULTS: Based on the TLFB, 118 subjects (22.8%) were heavy drinkers. The areas under receiving operating characteristics curves (AUROCs) were equivalent (≥0.898) for all questionnaires. When using the standard cut point of ≥8 for the AUDIT, the sensitivity was 0.48. Lowering the cut point to ≥5 led to both a sensitivity and specificity over 0.85. The optimal cut point of the AUDIT-C was ≥4. The AUDIT-QF, AUDIT-3 and elderly specific AUDIT-3 did not provide optimal combinations of sensitivity and specificity with any cut point. CONCLUSIONS: The AUDIT and AUDIT-C are accurate in screening for heavy drinking among the elderly if the cut points are tailored to this age group.