In an eastern population of song sparrows (Melospiza melodia), song type matching occurs at above chance levels but does not signal aggressiveness. One explanation for the apparent ineffectiveness of matching as a signal is that the occurrence of matching is constrained by internal rules for ordering the production of song types. This constraint hypothesis is tested here in an experiment in which the singing of territorial male song sparrows is monitored in the field in real time, and subjects are confronted with playback of one of their song types either immediately after switching away from that type (short delay) or after having cycled through much of their repertoire since last singing that type (long delay). Matching was not significantly more likely in the long delay treatment than in the short delay treatment. The probability of matching did, however, depend significantly on prior bout length: the longer was a singer’s last bout of a song type, the less likely the singer was to match it. There was also a suggestive effect of frequency of usage: males were more likely to match a song type the more frequently they normally sang that type, though this result was not significant after correcting for multiple comparisons. Thus, internal rules on how songs are sequenced exert constraints on the occurrence of song type matching, and such constraints can help to explain the apparent ineffectiveness of matching as a signal in this study population.