The yield of a cigarette is determined by the tobacco blend, the length of the cigarette, the cigarette paper, the filter and air dilution. Cigarette yield has been defined by tradition and by law to be the yield of nicotine, tar and carbon monoxide obtained from a 35 ml puff volume of 2-second duration taken every minute during the burning time of the cigarette. Normally smokers draw a puff into their mouth and then inhale. Mouth delivery is largely determined by personal smoking behaviour. The puff volume, number of puffs taken per cigarette, and number of cigarettes smoked per day determine both the volume and the mass of daily mouth delivery. There are marked differences in smoking behaviour, and the delivery is substantially altered from the yield values obtained with the standardised test procedure. Body uptake of smoke ingredients is determined by smoke chemical parameters, smoker inhalation behaviour, lung morphology, and physiological parameters. The physiological parameters include tidal volume, vital capacity, rate of breathing, and rate of clearance for the lung. Given these behavioural and physiological differences in individual delivery and uptake it is not surprising that differences in measured parameters occur within smokers of cigarettes with a particular yield. Biological differences among individuals, such as metabolic and size differences, cause additional variations in these values. Therefore, the estimates of nicotine and tar delivery can vary widely in studies of individual uptake when the estimates are based upon sample population data. The variables in both smoking behaviour and in chemical and physiological factors which alter uptake make it essential to have a crossover design for any study. The large standard error for the plasma concentration of cotinine (a major metabolite of nicotine) within a sample population, and the log linear nature of the plasma cotinine concentration curve, requires a very large sample size for any study of cigarette delivery or uptake. When comparisons of brands are made, average values are misleading in that the skew to the high values obscures frequency differences among the lower values within the samples. It is important to remember that smoker compliance with study design is very essential.(ABSTRACT TRUNCATED AT 400 WORDS)