Here, we tested for an ecologically valid cost to secondary sexual trait expression in the male great tit (Parus major). We compared the breast stripe area of a sample of males who survived extremely low night time temperatures with a sample who did not survive. Breast stripe area was significantly smaller in surviving males, suggesting a real cost of signalling in terms of survival. The relationship between breast stripe area and survival was moderated by the effects of body condition: Males with larger breast stripes were in poorer condition and hence suffered increased mortality. Finally, we tested for relationships between breast stripe area and body condition, and tarsus asymmetry and immunological parameters (Brucella abortus antibody count, heterophil to lymphocyte ratio, eosinophil count, monocyte count) in surviving males. Only eosinophil count was related to breast stripe area. Our results suggest a survival cost to investment in sexual signals, such that males who invest somatic resources in social signalling risk increased mortality under extreme environmental conditions.