Alcohol drinking is a risk factor for cancer. The degree of risk is increased in subjects showing the flushing response, which is due to aldehyde dehydrogenase 2 (ALDH2) polymorphism. The attributable risk of alcohol drinking for cancer in Japan has not been sufficiently investigated with consideration of flushing response. We followed 78,825 Japanese in JPHC study cohort II. The association between alcohol consumption and cancer incidence was assessed according to self-reported flushing response using Cox proportional hazards regression models. The population-attributable fraction (PAF) of cancer incidence was also estimated. During 1993-2013 (average follow-up, 16.8 years), 8486 incident cancers (included 4386 alcohol-related cancers) were reported. Half of men and 36% of women had flushing response. In men with flushing response, moderate or higher alcohol drinking increased the risk of alcohol-related cancers compared with non-drinkers (150-299 g/week, HR 1.63, 95% CI, 1.43-1.99; 300-449 g/week, HR 2.02 95% CI, 1.67-2.44; ≥450 g/week, HR 1.75, 95% CI, 1.39-2.21). When flushing response was considered in comparisons between non-drinkers and non-flushers, non-flushing heavy drinkers had a slightly higher PAF than flushers (all cancers: flushers, 2.0% and non-flushers, 2.2%; alcohol-related cancers: flushers, 3.8% and non-flushers, 5.8%). Although the risk of alcohol-related cancer in men with flushing response increased with increasing alcohol consumption, heavy drinkers were also at high risk regardless of flushing response. Considering the PAF of alcohol consumption on cancer, efforts to discourage heavy alcohol consumption to reduce the incidence of alcohol-related cancers appear warranted regardless of flushing response.