This report describes a new method for estimating the retention of selected mainstream smoke constituents in the respiratory tract of adult smokers during cigarette smoking. Both particulate-phase (PP) constituents including nicotine, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), and N'-nitrosonornicotine (NNN), two tobacco-specific nitrosamines (TSNA), and gas-vapor-phase (GVP) constituents including carbon monoxide (CO), isoprene (IP), acetaldehyde (AA), and ethylene, were studied. To estimate the amounts of smoke constituents delivered during smoking, we used predetermined linear relationships between the measured cigarette filter solanesol content and machine-generated mainstream deliveries of these selected compounds. To determine the amounts of smoke constituents exhaled, the expired breath was directed through a Cambridge filter pad (CFP) attached to an infrared spectrometer. PP compounds were trapped on the CFP for later analysis and GVP compounds were analyzed in near real time. The smokers' respiratory parameters during smoking, such as inhalation/exhalation volume and time, were monitored using LifeShirt(R), a respiratory inductive plethysmography (RIP) device. The retention of each smoke constituent, expressed as a percentage, was then calculated as the difference between the amount delivered (estimated) and the amount exhaled relative to the amount delivered. We studied 16 adult male smokers who smoked cigarettes according to 3 predefined smoking patterns: no inhalation (pattern A), normal inhalation (pattern B), and deep inhalation (pattern C). For the three PP constituents, the mean retentions for pattern A ranged between 10 and 20%; and while the mean retentions of the two TSNAs were significantly higher for pattern C (84% for NNK and 97% for NNN) than those for pattern B (63% for NNK and 84% for NNN), the mean retentions of nicotine were basically the same between patterns B and C, which were both greater than 98%. For the GVP constituents, the retentions were similar between pattern B and pattern C, although different constituents were retained to different degrees (average values of 33%, 52%, 79%, and 99% for ethylene, IP, CO, and AA, respectively). The differences in the retention between different constituents could be interpreted in terms of each constituent's physical properties such as volatility and solubility. In conclusion, the method described is suitable for studying the retention of selected mainstream smoke constituents in the respiratory tract of smokers.