We report an experiment in which participants ( N = 368) were asked to differentiate between major and minor thirds. These intervals could either be formed by diatonic tones from the C major scale (tonal condition) or by a subset of tones from the chromatic scale (atonal condition). We hypothesized that in the tonal condition intervals would be perceived as a function of scale step distances, which we defined as the number of diatonic leaps between two notes of a given music scale. In the atonal condition, we hypothesized that intervals would be perceived as a function of cents. If our hypotheses were supported, we should verify a less accurate performance in the tonal condition, where scale step distances are the same between major and minor thirds. The data corroborated our hypotheses, and we suggest that acoustic measurements of intervallic distances (i.e., frequency ratios and cents) are not optimal when it comes to describing the perceptual quality of intervals in a tonal context. Finally, our research points to the possibility that, in comparison with previous models, scale steps and cents might better capture the notion of global versus local instances of auditory processing.