Most biomarkers of exposure tend to have short half-lives. This includes cotinine, a metabolite of nicotine widely used to assess smoke exposure. Cotinine is thus unsuitable as a determinant of past exposure to cigarette smoke. We used bisulphite pyrosequencing of a set of four genomic loci (AHRR, 6p21, and two at 2q37) that had differential DNA methylation levels in peripheral blood DNA dependent on tobacco exposure to create a predictive model of smoking status. Combining four gene loci into a single methylation index provided high positive predictive and sensitivity values for predicting former smoking status in both test (n = 81) and validation (n = 180) sample sets. This study provides a direct molecular measure of prior exposure to tobacco that can be performed using the quantitative approach of bisulphite pyrosequencing. Epigenetic changes that are detectable in blood may more generally act as molecular biomarkers for other exposures that are also difficult to quantify in epidemiological studies.