We developed a nonobtrusive technique to measure the volume puffed during smoking. An inductive plethysmographic coli placed around the cheeks was calibrated by coaching the subject to take graded puffs from an unlit cigarette connected in series to a 2-L Krogh spirometer. The device had a linear volume response that was unaltered by smoking filtered or unfiltered cigarettes or by the length of the cigarette. It remained calibrated after smoking 4 cigarettes over 150 min. Ten subjects smoked a high tar content (HTC) and a low tar content (LTC) cigarette, each with and without a cigarette holder. The mean puff volume from LTC of 52 +/- 15 ml was significantly higher than HTC of 39 +/- 10 ml (p less than 0.001). The number of puffs per cigarette (8.4 +/- 2.1 for LTC, 8.3 +/- 2.0 for HTC), the volume of inhaled (841 +/- 517 ml for LTC and 748 +/- 323 ml for HTC), and the duration of inhalation and associated breathhold (5.5 +/- 2.6 s for LTC and 5.6 +/- 2.4 s for HTC) as measured by respiratory inductive plethysmography were comparable. The perception of intensity of inhalation by the subjects bore little relationship to the objective measurement. A cigarette holder increased the number of puffs per cigarette to 12.6 +/- 4.8 for LTC (p less than 0.02) and to 11.9 +/- 3.6 for HTC (p less than 0.01 and increased the puff volume for HTC to 50 +/- 12 ml (p less than 0.01) and the duration of inhalation to 6.6 +/- 2.5 s for LTC (p less than 0.005). These results show that (1) smokers switched acutely from HTC to LTC draw larger puffs from LTC thereby possibly circumventing the perceived benefits of smoking LTC, (2) cigarette holders modify smoking pattern and hence if used with an incorporated flowmeter to monitor the natural pattern of smoking might give spurious information.