Eight short-term (24–48 h) air samples were taken at Lancaster, UK, to study the gas–particle partitioning of PCDD/Fs. Sampling dates in autumn 1997 were selected with a view to minimising temperature fluctuation during the sampling events. ΣCl4-8DD/Fs (ΣTEQ) for the first 6 samples were 1.1–3.6 pg m−3 (15–44 fg TEQ m−3), typical of a rural site; two other samples had ΣCl4-8DD/Fs of 18 and 7.9 pg m−3, with 320 and 100 fg TEQ m−3. The observed gas–particle distributions varied from 0–34% particle-bound for Cl2/3DD/Fs to >70% for Cl6-8DD/Fs. Measured particle-bound fractions were compared to theoretical estimates of their distribution based on the Junge–Pankow model using three different reported sets of vapour pressures. The best correlation was obtained using vapour pressures derived from measured GC-retention time indices (Eitzer and Hites, 1988). Plotting log partition coefficient (Kp) versus log sub-cooled liquid vapour pressure (pL) gave excellent correlations with slopes of roughly −1 for all homologue groups. 2, 3, 7, 8-substituted congeners showed slopes of −1 for the first five sampling events. It is proposed that kinetic factors at the low ambient temperatures, coupled with additional emissions during the last sampling events resulted in non-equilibrium partitioning.