Two experiments were conducted to investigate the perception of structural boundaries in six popular music songs. In the segmentation experiment, participants were asked to indicate perceived segment boundaries in monophonic representations of the songs, synthesized from the MIDI score. In the salience rating experiment, participants were asked to rate the salience of a number of boundaries selected from the outcome of the segmentation experiment, and to describe the perceptual cues for each boundary. The segmentation experiment showed that there is a wide variety in the number and temporal positions of perceived boundaries across participants. However, certain boundaries in the music are indicated by nearly all participants. The salience rating experiment showed a moderate correlation between participants’ boundary salience ratings. Comparing the outcome of the two experiments, we found a significant correlation between the frequency of boundary indications and the corresponding salience rating of that boundary. These findings suggest that both methods can be used equally well for evaluating the perceptual boundaries. The perceptual boundaries were also compared to boundaries predicted by three musicological models. The comparison of the perceptual boundaries with the predicted boundaries showed a moderate correlation between the perceptual and predicted boundaries.