There is growing evidence for the role of smoking in the aetiology of multiple sclerosis. We have expanded existing meta-analyses and further explored the roles of study design, gender, latitude and year of study with regression modelling. We have found a consistent association between smoking and MS with an odds ratio of approximately 1.5, with males at higher risk. This finding is independent of study design. However, latitude and year of study may have unexpected influence. Smoking appeared to confer a greater risk to females living closer to the equator than to females at higher latitudes. The effect of cigarette smoke exposure on MS risk may not be fixed over time, but could be increasing. These results suggest a threshold model of MS risk that includes a fairly constant genetic risk (for Caucasian populations) together with variable environmental risks which are dominated by vitamin D deficiency at higher latitudes and are more significant in women who have an intrinsically lower threshold for development of disease.