Relationships between machine smoking nicotine yield and different smoke exposure indicators were investigated in a cross-sectional study. For each of the four yield classes H (1.0-1.2 mg), M (0.7-0.9 mg), L (0.4-0.6 mg) and U (0.1-0.3 mg) 18 male and 18 female subjects were recruited. The experimental design (2 x 2) included smoking with lip contact or with a flowmeter holder, natural smoking of one cigarette or forced smoking (30 puffs). The analysis of presmoking measures revealed for plasma nicotine H greater than L, U; M greater than U, for plasma cotinine H, M greater than U, and no differences for respiratory CO. Pre- to postsmoking boosts of CO and nicotine increased with yield, but the differences were smaller than those in yield. This partial compensation can be attributed to puffing behavior as revealed by the differences between yield classes with respect to flowmeter measures (puff volume, flow parameters, number of puffs). Contact condition hardly influenced the results. Forced puffing revealed down regulation mechanisms in smoke absorption and, less pronounced, in puffing behavior. Cardiovascular and subjective effects were widely independent of yield. Plasma cotinine appeared as the best smoke exposure indicator, due both to its high retest reliability and its relationship to nicotine yield.