Low amplitude signals function in private exchanges of information between signalers and nearby receivers. The eavesdropping avoidance hypothesis proposes that selection favors quiet threat signals in order to avoid the costs of eavesdroppers. If true, then selection should favor other acoustic traits in addition to low amplitude that lead to quiet signals transmitting less effectively through the environment compared to broadcast signals. The “warbled” soft songs of male song sparrows differ from “crystallized” soft songs and from broadcast songs in a number of acoustic traits, suggesting that these songs may transmit less effectively. We tested this prediction in a field experiment by playing broadcast songs, crystallized soft songs, and warbled soft songs through a loudspeaker at the same amplitude and recording the propagated songs at five distances, at two heights, and in two different habitat types. Counter to our prediction, we found no evidence that either form of soft song transmits differently than broadcast song when all were played loudly. If anything, soft songs transmitted more effectively when all songs were played quietly. Our results do not support one prediction made by the eavesdropping avoidance hypothesis, although the possibility remains that reduced amplitude alone is sufficient to reduce eavesdropping. The question of why warbled soft song differs in acoustic structure remains unresolved.