Eight conventional U.S. market cigarettes, representing a cross-section of design parameters, two blended reference cigarettes, and an electrically heated prototype cigarette (EHC) were machine-smoked under US Federal Trade Commission conditions and Massachusetts DPH conditions. Smoke was evaluated for chemical composition and in vitro toxicity. Theoretical risk estimates were calculated using a model based on the U.S. EPA approach for chemical mixtures. Regression curves were fitted to the data from the market cigarettes to investigate potential relationships. Puffing profiles for two of the market cigarettes and the EHC were determined in humans; smoke generated under these conditions was evaluated for the same endpoints. Generally, the yields of individual smoke constituents, in vitro toxicity, and theoretical risk estimates per cigarette increased with increasing yields of total particulate matter (TPM), but the increase was less than proportional. These relationships could be described well by non-linear regression. For example, when increasing the TPM yield by a factor of 4 (range: 5 to 20 mg/cigarette), the bacterial mutagenicity in strain ta100 increased by a factor of 3.1, i.e., 22% less than proportional. On average, the increases in smoke constituent yields, in vitro toxicity, and theoretical risk estimates were 15% less than proportional in this range, irrespective of whether the increased TPM yields were due to smoking conditions or cigarette construction. This analysis indicates that at the same TPM yield, the smoke constituent yields, in vitro toxicity, and theoretical risk estimates per cigarette were similar for low and high yield cigarettes. The two market cigarettes, when machine-smoked using human puffing profiles, and the reference cigarettes fitted fairly well to these relationships, supporting the data interpretation. The EHC was consistently approximately 90% lower in all endpoints as regression analysis for conventional cigarettes would suggest.