Around the world, international machine smoking norms (e.g. ISO and FTC) are used to determine tar and nicotine yields of cigarettes, with the main purpose of comparing smoke yields from the cigarettes. There has recently been an expectation by some authorities that smoke yields determined by standard machine methods should correspond to yields obtained from the cigarette by individual human smokers. This is unrealistic because of the variability in human smoking conditions. The aim of this study was to assess the smokers' tar and nicotine yields under natural smoking conditions, and to compare them with the ISO yields. We asked 61 usual smokers of two main commercial brands of regular cigarettes (12 mg tar and 0.9 mg nicotine) in France to participate in the study and to collect 5 of the butts from cigarettes that they had smoked under natural conditions. The smoking behaviour of each subject was determined using a Smoking Puff Analyser. Duplication of each smoking profile was then undertaken in order to determine the tar and nicotine yields obtained under these laboratory conditions. In the past, many of the smoking behaviour studies that have been undertaken to measure the true representation of human smoking yields generally stopped here. But it is easy to prove that subjects are perturbed by laboratory conditions, resulting in mean increases in smoke yields of 40% compared to natural smoking conditions. Because high correlations have been observed between the optical density of alcoholic solutions prepared from the butts and the amount of tar and nicotine produced by puff duplication, we used the first laboratory optical density results to evaluate the smoke yields under natural conditions. Both groups of smokers produced an average of 12.1 mg tar and 0.92 mg nicotine, with large differences between smokers. We conclude that the standard smoking machine conditions give a good indication of mean tar and nicotine yields amongst smokers for the 12 mg tar cigarettes used in this study.