This study examined the identification of English consonants presented in noise. The subjects were English native (NE) speakers and four groups of Italian/English bilinguals who differed according to age of arrival (AOA) in Canada and/or self-estimated percentage daily use of Italian (%ITAL). The naturally produced stimuli, which differed according to their initial or final consonant (e.g., ``bado,'' ``sado,'' ``hodab,'' ``hodas''), were presented at decreasing S/N ratios (+12, +6, 0, -6 dB). Subjects identified each consonant by selecting one of five written alternatives, including the target and four likely foils. Percent correct identification scores decreased similarly across the four S/N ratios for all five groups. However, the bilinguals scored lower than did the NE subjects, despite their long residence in Canada (M=35.4 years). Their scores varied inversely with AOA and %ITAL, and were related to their performance on a test of phonological short-term memory. Which consonants were most difficult to identify varied across the five groups. Error analyses revealed that errors due to confusions of place and manner of articulation were similar across the five groups. Additional analyses will determine whether the influence of the L1 can be characterized in terms of finer-grained features.