Urbanization changes the physical environment of non-human species, but also markedly changes their acoustic environment. Urban noise interferes with acoustic communication in a range of animals including birds, with potentially profound impacts on fitness. However, a mechanistic theory to predict which species of birds will be most affected by urban noise is lacking. We develop a mathematical model to predict the decrease in the active space of avian vocal signals when moving from quiet forest habitats to noisy urban habitats, and find that the magnitude of the decrease is largely a function of signal frequency. However, this relationship is not monotonic. A meta-regression of observed increases in the frequency of birdsong in urban noise supports the model's predictions for signals with frequencies between 1.5 and 4 kHz. Using the results of the meta-regression and the model described above, we show that the expected gain in active space following observed frequency shifts is up to 12%, and greatest for birds with signals at the lower end of this frequency range. Our generally-applicable model, along with three predictions regarding the behavioral and population-level responses of birds to urban noise, represents an important step towards a theory of acoustic communication in urban habitats.