It has been proposed that low-frequency oscillations are involved in separating neuronal representations belonging to different stimulus items. Indeed, item-specific neuronal activity was found to cluster on different oscillatory phases. However, the consequences of this neural mechanism for perception are unknown. Here, we investigated whether and how neuronal item separation through oscillatory clustering influences perceptual item separation. In an EEG experiment, participants categorized sounds parametrically varying in pitch, relative to an arbitrary pitch boundary. We found that pre-stimulus oscillatory phase, in the theta and alpha ranges, biased near-boundary sound categorization responses to one category or the other. Phase also modulated whether evoked neuronal responses more closely resembled the sound envelope of one or another category. Intriguingly, participants with stronger oscillatory clustering (phase strongly biasing sound categorization) in the theta, but not alpha range, had steeper perceptual psychometric slopes (sharper discrimination between sound categories). These results indicate that neuronal sorting of information by phase directly influences subsequent perception, and has a positive impact on discrimination performance.