Traits that influence social interactions (signals) are widely thought to be honest indicators of some underlying quality of their bearer. One hypothesis is that a signal and the bearer's immunocompetence are linked via a condition-dependent pathway. We tested this idea by measuring the effect of specific dietary components on the development of a melanin-based plumage signal and the production of antibodies in juvenile male house sparrows entering their first moult. We maintained sparrows on one of three artificial diets: high total protein, low total protein, or intermediate protein with the precursors of melanin, phenylalanine and tyrosine (PT), reduced by 50%. Diet did not affect survival or weight gain. Two aspects of male plumage differed between treatments; the white wing bar was significantly smaller in low-protein males, and the black bib feathers had significantly higher reflectance in PT-reduced males. PT reductions had no effect on bib size. PT-reduced birds also produced more antibodies to diphtheria and tetanus antigens than did other subjects. After repeating the experiment using a better control and a different diet formula to compensate for an energy imbalance resulting from reduced PT levels, PT reduction again produced lighter bib feathers, but the effect of diet on antibody production disappeared. We conclude that the amino acid precursors to melanin affect melanin synthesis if scarce in the diet, but in this case, do not affect size, the most conspicuously variable aspect of the signal. We found no evidence of a condition-dependent link between melanin synthesis and immunocompetence.