Background Alcohol consumption is pivotal for the subsequent development of alcohol use disorders (AUD). The Alcohol Use Disorders Identification Test (AUDIT) is a recommended AUD screening tool for prevention and primary care settings. The objectives of this study were to test how many participants with heavy drinking are unidentified by the AUDIT, if proportions of unidentified participants vary over time, and whether this unidentified risk group (URG) was clinically relevant in terms of drinking behavior reports and AUD risk factors, as well as future adverse outcomes, such as craving, dependence symptoms, or depression. Methods Our prospective cohort study followed 164 German males aged 18–19 years without an alcohol dependence diagnosis over 24 months. Only men were included due to higher AUD prevalence and gender-specific differences in metabolism, drinking patterns, and progression to AUD. All participants were screened via telephone interview and answered questionnaires both in person and via internet. Heavy drinking was classified using the AUDIT consumption score (AUDIT-C ≥ 4.50). Standardized AUD diagnoses and symptoms, as well as alcohol use-related outcome criteria were assessed via standardized Composite International Diagnostic Interview (CIDI), and self-report questionnaires. Results One in four participants (22–28% across all four follow-ups) reported heavy drinking but was unidentified by AUDIT total score (i.e. score < 8), thus qualifying for URG status. The URG status did not fluctuate considerably across follow-ups (repeated-measures ANOVA, p = 0.293). URG participants identified at the six-month follow-up did not generally differ from participants without URG status in terms of AUD family history or temperament (multivariate ANOVA, p = 0.114), except for anxiety sensitivity (p_Bonferroni < 0.001). After two years, URG participants reported a similar level of adverse outcomes compared to low-risk participants (multivariate ANOVA, p = 0.438), but less alcohol-related problems and less loss of control due to craving compared to high-risk participants (p_Bonferroni ≤ 0.007). Conclusions Despite the considerable number of heavy-drinking individuals unidentified by AUDIT total scores, an additional classification according to AUDIT-C values did not prove useful. Combining AUDIT and AUDIT-C scores might not be sufficient for identifying AUD risk groups among young adult German males. There is an urgent need for a replication of our findings among female participants.