Previous perceptual studies of English stop voicing focus on Voice-Onset Time (VOT). Aspiration is generally subsumed into VOT, yet [1] complicates this, evincing a trading relation between intensity of aspiration noise and VOT. Our study is the first to examine the role of VOT and aspiration in English listeners’ perception of non-native plain and murmured stops. We recorded Marathi talkers producing /CVsV/ nonce words beginning with /t/, /tʰ/, /d/, /dʰ/, e.g. /dʰaːsaː/, or their velar counterparts. The following acoustic measures were taken for each token: — Duration of prevoicing — After Closure Time (ACT) [2], i.e., the interval between release and periodicity — Pre-Vocalic Interval (PVI) [3], which includes ACT and the murmured portion of the vowel — Intensity of aspiration noise Canadian English listeners rated the stops on a 6-point scale from voiced to voiceless. Only results for murmured stops are discussed here. Despite variability of prevoicing duration in these tokens (range = 200ms, sd = 43), the factor did not correlate significantly with a token’s average rating (p>0.60). However, ACT (r=.53, p<0.001), PVI (r=0.39, p<0.001), and the mean intensity of aspiration noise (r = 0.58, p<0.02) did. Thus, aspiration, not prevoicing, best accounts for perceptual differences between murmured stops. [1] Repp, “Relative amplitude of aspiration noise…,” Lang. Speech 22, 1979; [2] Mikuteit & Reetz, “Caught in the ACT…,” Lang. Speech 50, 2007; [3] Berkson, “Capturing breathy voice…,” Kans. Work. Pap. Ling. 33, 2012.