The aim of this study was to determine the relative importance of source parameters (fundamental frequency and waveform) versus formant parameters (resonant frequency and number) in identifying meaningless complex sounds. Sixteen sounds were generated differing on four dimensions, each having two values. These were the source waveform, containing either all harmonics or odd harmonics only; the fundamental frequency, either 90 or 142 Hz, ratio=1.6; the number of formants, one or two; and the frequency of the formant(s) either low (600 Hz, or 600 and 1550 Hz) or high (940 Hz, or 940 and 2440 Hz) high/low ratio = 1.6. Twenty-four listeners were trained to identify these sounds. The results show that fewer confusions were made between pairs of sounds (1) as the number of dimensions on which they differed increased, and (2) as the dimension (single or in combination) changed, from formant number to formant frequency region, to fundamental frequency, to source waveform. Listeners appear to make a generalized classification along a single dimension which seems to vary systematically with waveform complexity and periodicity.