To compare the effect of potentially modifiable lifestyle factors on the incidence of vascular disease in women with and without diabetes. In 1996-2001 over one million middle-aged women in the UK joined a prospective study, providing medical history, lifestyle and socio-demographic information. All participants were followed for hospital admissions and deaths using electronic record-linkage. Adjusted relative risks (RRs) and incidence rates were calculated to compare the incidence of coronary heart disease and stroke in women with and without diabetes and by lifestyle factors. At recruitment 25,915 women (2.1% of 1,242,338) reported current treatment for diabetes. During a mean follow-up of 6.1 years per woman, 21,928 had a first hospital admission or death from coronary heart disease (RR for women with versus without diabetes = 3.30, 95% CI 3.14-3.47) and 7,087 had a first stroke (RR = 2.47, 95% CI 2.24-2.74). Adjusted incidence rates of these conditions in women with diabetes increased with duration of diabetes, obesity, inactivity and smoking. The 5-year adjusted incidence rates for cardiovascular disease were 4.6 (95% CI 4.4-4.9) per 100 women aged 50-69 in non-smokers with diabetes, 5.9 (95% CI 4.6-7.6) in smokers with diabetes not using insulin and 11.0 (95% CI 8.3-14.7) in smokers with diabetes using insulin. Non-smoking women with diabetes who were not overweight or inactive still had threefold increased rate for coronary disease or stroke compared with women without diabetes. Of the modifiable factors examined in middle aged women with diabetes, smoking causes the greatest increase in cardiovascular disease, especially in those with insulin treated diabetes.