AIMS: Very few studies indicating that low-moderate alcohol consumption protects from myocardial infarction (MI) controlled for social support and working conditions, which could confound the findings. Therefore, a first aim was to study the risk of non-fatal and total MI in relation to volume of alcohol consumption and measures of social support and working conditions. A second aim was to analyse the impact of the volume of earlier alcohol use in abstainers. DESIGN: Data came from a case-control study, the Stockholm Heart Epidemiology Program (SHEEP), including first MI among Swedish citizens 45-70 years old. SETTING: Stockholm County 1992-94. PARTICIPANTS: There were 1095 cases of MI in men and 471 in women (928 and 372 were non-fatal), and 2339 living controls from the general population. MEASUREMENT: Information about alcohol use at different periods in life and job strain, social anchorage and life control besides pre-existing health problems, smoking, physical activity, socio-economic status and marital status was obtained by a questionnaire from the cases and the controls. FINDINGS: In multivariate logistic regression analyses, the relative risk for MI (especially non-fatal) was reduced among alcohol consumers. RR for non-fatal MI was 0.52 (95% confidence intervals 0.32, 0.85) in men with a consumption of 50-69.9 g 100% ethanol/day and 0.21 (95% confidence interval 0.06, 0.77) in women with a consumption of 30 g or more per day (reference category 0.1-5 g 100% ethanol/day). Men who were abstainers during the previous 1-10 years and with an earlier average consumption of 5-30 g 100% ethanol/day had a significantly lower relative risk compared to such abstainers with an earlier higher consumption. Earlier consumption among abstainers may also have an impact on gender differences in MI. Analyses showed positive interaction between abstention and low life-control in women, but only 4% of the female cases were due to this interaction. There were no other interactions between measures of alcohol use and social anchorage, life control and working situations. CONCLUSION: Alcohol use had a protective impact on MI, with little impact of job strain, social anchorage and life control, giving increased support for a protective impact of low-moderate alcohol use. The level of previous alcohol consumption among male 1-10-year-long abstainers influenced the risk of MI.