BACKGROUND: Identifying dominant processes that underlie the development of other processes is important when evaluating the temporal sequence between disorders. Such information not only improves our understanding of etiology but also allows for effective intervention strategies to be tailored. The temporal relationship between alcohol intake and mental health remains poorly understood, particularly in nonclinical samples. The purpose of this study was to disentangle the dominant temporal sequence between mental health and frequency of heavy drinking days. METHODS: We report a 10-year (1997 to 2007) prospective cohort study of 500 respondents (74% male) from the Birmingham Untreated Heavy Drinkers project. Participants were aged 25 to 55 years at baseline, drinking a minimum of 50/35 U.K. units of alcohol for men/women on a weekly basis, and were not seeking treatment for their alcohol use upon recruitment into the study. Heavy drinking days were defined as consuming 10/7+ U.K. units of alcohol in a single day for men/women. Mental health was assessed using the mental health component score of the SF-36 questionnaire. Dynamic longitudinal structural equation models were used to test competing theoretical models (frequency of heavy drinking days leading to changes in mental health scores and vice versa) and a reciprocal relationship (both mental health scores and the frequency of heavy drinking days influencing changes in each other). RESULTS: A model whereby mental health scores were predictors of change in the frequency of heavy drinking days was of best fit. In this model, mental health scores were negatively related to change in heavy drinking days (β -0.80, SE 0.28) indicating that those with higher mental health scores (i.e., better functioning) made larger reductions in the number of heavy drinking days over time. CONCLUSIONS: Mental health appears to be the stronger underlying process in the relationship between mental health and frequency of heavy drinking days.