Blood nicotine, cotinine, and carboxyhaemoglobin (COHb) concentrations were measured in 392 smokers (255 women and 137 men) of "middle tar" (17-22 mg), "low to middle" (11-16 mg), and "low tar" (less than 11 mg) cigarettes. Since tar intake cannot yet be measured directly, we devised an index to estimate it based on the use of measured levels of an intake marker (eg, blood nicotine) and the ratio of the tar to marker yields of the cigarettes. This approach was validated by its ability to enhance the prediction of levels of one marker by use of another. In a practical test, using COHb and the CO/nicotine yield ratio of the cigarettes, the mean blood nicotine concentration of the low tar smokers was predicted to be 31.9 ng/ml compared with the measured mean of 31.8 ng/ml. Our main findings were that despite substantial compensatory increases in inhalation, the low tar smokers took in about 25% less tar, about 15% less nicotine, and about 10% less carbon monoxide than smokers of middle and low to middle tar cigarettes. These results indicate that low tar cigarettes of the type available in Britain since the late 1970s are likely to prove less harmful than other brands. Monitoring of smoke intakes could supplement epidemiological approaches and provide earlier evidence of whether changing cigarette designs lead to any significant dosage reduction that could affect the risk of disease.