In order to investigate the effect of level of presentation on subjects’ labeling of speechlike sound patterns, synthetic stimuli were constructed, varying systematically F0contours, VOT, and F2transitions. These stimuli were presented in random sequences at levels between 15 and 105 dB SPL to subjects with normal hearing. No significant response variation was observed in the range 40–100 dB SPL. Subjects’ labeling behavior suddenly breaks down below levels of around 35 dB SPL. The secondary findings of the study are also discussed in terms of different specific processing strategies for different specific speech features.