Tracer gas concentrations were measured on a 60%-sized mannequin holding a pure sulfur hexafluoride source in its hands at waist height while it stood in a wind tunnel. Samplers were placed at the mannequin's mouth, in front of the ear, and at three chest locations at lapel level. Simultaneous 15-min time-weighted average samples were taken by drawing air into different sampling bags with sampling pumps. For the factorial study design, test conditions included cross-draft velocities of 10, 22, 47, and 80 ft/min; three mannequin orientations (facing to, side to, and back to cross-draft), and rotating speed through an 80 degrees arc (fast, slow, and no movement). Each study condition was tested twice. Concentrations at all sampling locations when the mannequin faced to the front and side were less than a tenth of the levels measured at the nose (Cnose) when the mannequin faced downstream. Higher velocities significantly increased concentration at the Back orientation and generally reduced it at the Side and Facing orientations. Concentrations at the nose were different from concentrations at other sites. For 34 of 36 samples the mean chest concentration (Cchest,) was higher than the Cnose (geometric mean three times higher). The ratio of ear (Cear) and Cnose varied with orientation. At the Back orientation, Cear, was lower than Cnose, whereas Cear was higher than Cnose at the Side and Facing to flow orientations. Velocity affected the ratios of concentrations. At the Back orientation, the chest sampler provided lower overestimates of Cnose, at higher velocities than at lower values. Mannequin movement, done only at the Back orientation, proved important only for the ear location. Results showed significant and substantial differences between concentrations at the nose and lapel. However, these findings should be interpreted with caution because a very dense tracer gas and an unheated, nonbreathing mannequin were used. In more realistic conditions, the findings probably would show far smaller differences in concentrations at different sampling sites.