This note is concerned with recent research into memory for tunes and tonal sequences, and is intended to show that there are serious conceptual problems in this area. Reference is made to work by Dowling & Fujitani (1971), Cuddy & Cohen (1976), and to an earlier paper by Deutsch (1969). It is not concerned with more widely studied problem of memory for individual tones. Experimental work on memory for tunes has relied heavily on recognition paradigm; and/or upon use of short (two- or three-tone) sequences. It is not clear that conclusions derived from recognition unambiguously point to an interval abstraction process; nor that use of extremely short stimulus sequences throws any light on memory for real melodies. Evidence exists which suggests that truly melodic sequences are not remembered in terms of interval-magnitudes, and that consequently postulation of a neurological model for such a process may be premature.