This paper compares the production and perception of the rise-fall contour of contrastive statements and the final rise-fall part of polar questions in Greek. The results show that these superficially similar rise-falls exhibit fine phonetic differences in the alignment of tonal targets with the segmental string, and that these differences can be used by native speakers under experimental conditions to identify the two contour types. It is further shown here that the observed differences in alignment are best attributed to differences in the overall tonal composition of these contours, which results in different degrees of crowding for the targets involved. This analysis accounts for the differences in phonetic detail between the two contours, while obviating the need to posit distinct secondary associations for the peak of the rise-fall. It is suggested that differences in phonetic alignment should be formalized by means of the secondary association mechanism only if simpler analyses and explanations have been considered and shown not to account effectively for the data. Finally, the perceptual results suggest that even small alignment differences like those observed here have a role in perception and should therefore be specified in a full description of the phonetic implementation of tunes.