A single even harmonic added to an odd-harmonic complex is often judged to be more salient than its odd neighbors in a clarity rating task [Roberts and Bregman, J. Acoust. Soc. Am.90, 3050–3060 (1991)]. This study used similar complexes in a two-interval forced-choice procedure. Each interval consisted of a complex tone followed by a pure tone, whose frequency matched that of a harmonic in one interval but was changed by ±0.5×fundamental frequency in the other. Subjects were asked to identify the matching interval. Since the pure tone followed the complex tone, it could not cue listening to a particular frequency region. The possibility of cross-interval cuing was reduced by changing the fundamental frequency between intervals (100–150 Hz range). The procedure was designed to maximize the effects on performance of differences in immediate perceptual salience between the partials. The added even harmonic was typically judged with greater accuracy than its odd neighbors (experiment 1), though this effect was greatly reduced for harmonics above 8 (experiment 2). The even–odd difference persisted when the original stimuli were made inharmonic by applying a frequency shift of 15%, but was abolished for stimuli consisting of successive partials (experiment 3).