This study describes selection derived from habitat acoustics on the physical structure of avian sounds. Sound propagation tests were made in forest, edge, and grassland habitats in Panama to quantify pure tone and random noise band sound transmission levels. The sounds of bird species in each habitat were analyzed to determine the emphasized frequency, frequency range, and sound type (whether pure tonelike or highly modulated). Forest habitats differ from grass and edge in that a narrow range of frequencies (1,585-2,500 Hz) has lower sound attenuation than lower or higher frequencies. Attenuation increases rapidly above 2,500 Hz. Bird sounds from species occurring at the lower forest levels were found to be predominantly pure tonelike with a frequency emphasized averaging 2,200 Hz, conforming to the predictions based on sound propagation tests. The edge habitat is characterized by a wide range of frequencies having a generally similar attenuation rate. Pure tone and random noise band sounds did not differ in attenuation rates. The sounds of edge birds have high variance in the frequency-emphasized component, and more are composed of both modulated and pure tone elements to a greater extent than forest or grassland species. It is suggested that this variance is a reflection of a relative lack of selection pressure from habitat acoustics. Grassland propagation tests showed a positive correlation between increasing frequency and increasing attenuation. It is hypothesized that information transfer may be based on temporal aspects of the signal since these are less distorted in temperature- and wind-speed-stratified open environments.