During 826, 1-h sessions 107 participants (33 women, 74 men, 18-51 years of age) were exposed to defined draughts where air velocity (0.1 < or = V(a) < or = 0.4 m s(-1)), turbulence intensity (< 30, approximately equal to 50, > 70%), draught direction (horizontal, vertical, diagonal), air temperature (11 < or = t(a) < or = 23 degrees C), and metabolic rate ( approximately equal to 60< or = M < or = 156 W m(-2)) were varied but kept constant during single sessions. Perception and evaluation (annoyance due to draught and local cold) were registered using a list of prescribed body parts. Annoyance due to draught increased with the increase in mean air velocity and in turbulence intensity, as well as with the decrease in air temperature and in metabolic rate. The direction of air flow had no influence. As these responses were neither sufficiently predicted by the draught-rating model proposed in ISO 7730 (1994), nor by its extension developed by Toftum (1994a), the latter was modified according to the data registered in this paper. Using the modified model the accordingly predicted percentage of people annoyed by draught agreed highly with the percentage actually observed (r=0.726, p<0.001). As the resulting model covers larger ranges of air temperature and metabolic rate it is suggested that it should substitute the model proposed in ISO 7730 [1994].