The mechanism of cough associated with upper respiratory infection (URI) is poorly understood. This paper reports a study of the role of altered sensitivity of capsaicin-sensitive airway nerves. In a prospective study, baseline (B) capsaicin-induced cough and methacholine-induced airway responsiveness were measured in 103 healthy volunteers. During the following year, 31 subjects reattended for challenge testing during URI (I) and after recovery (R). The log concentration of capsaicin required to elicit two coughs (C2) was significantly lower during infection than recovery but not baseline [median (interquartile range) B = 0.59 (0.28-1.20), I = 0.27 (0-0.89), R = 0.89 (0.28-1.49)]. Log C5 (concentration causing five coughs) was lower during infection than baseline and recovery [B = 1.79 (1.20-2.70), I = 1.49 (0.89-2.08), R = 1.79 (1.20-2.40)]. FEV1 and PC15 methacholine values were unchanged during infection compared to baseline. Subjects with dry cough (n = 14) had lower C5 values during infection than both baseline and recovery, and lower C2 values during infection than recovery; in these subjects, increase in capsaicin sensitivity correlated with cough severity score. Subjects with productive cough or no cough showed no consistent changes during infection. Twenty-six control subjects who reattended without URI showed no change in capsaicin sensitivity. Upper respiratory infection may cause cough as a result of increased sensitivity of capsaicin-sensitive afferent airway nerves without affecting airway calibre or responsiveness.