The present investigation assessed the simultaneous and temporal masking produced by computer‐generated synthetic vowels. The durations (100 and 200 ms) of each of four vowel‐like maskers were employed. The masker was presented at 70 dB SPL. The probe signals were three filtered noise bursts whose spectral distributions corresponded to regions of high spectral energy in three English stop consonants. Quiet and masked thresholds were determined using the method of adjustment. Data are reported for two experienced listeners who participated in all the listening conditions. The results were generally in accord with the results of masking experiments using nonspeech signals in that both the frequency specificity of masking and temporal masking effects were demonstrated.