The question of whether musical scales have developed from a processing advantage for frequency ratios based on small integers, i.e., ratios derived from relationships among harmonically related tones, is widely debated in musicology and music perception. In the extreme position, this processing advantage for these so-called "natural intervals" is assumed to be inherent, and to apply to sequentially presented tones. If this is the case, evidence for this processing advantage should show up in psychoacoustic experiments using listeners from the general population. This paper reports on replications and extensions of two studies from the literature. One [Lee and Green, J. Acoust. Soc. Am. 96, 716–725 (1994)] suggests that listeners from the general population can in fact determine whether sequentially presented tones are harmonically related. The other study [Houtgast, J. Acoust. Soc. Am. 60, 405–409 (1976)] is interpreted in different terms, but could be confounded by such an ability. The results of the replications and extensions, using listeners of known relative pitch proficiency, are consistent with the idea that only trained musicians can reliably determine whether sequentially presented tones are harmonically related. © 1999 Acoustical Society of America.