Relatively few studies have dealt with the interaction between linguistic information and talker cues in speech perception and word recognition. The present study examines the effects of talker familiarity on serial recall for spoken word lists. Compared to lists of words produced by a single talker, lists produced by multiple talkers lead to decreased accuracy of serial recall in the primary portion of the recall curve at fast presentation rates. This pattern of recall for multiple‐ and single‐talker stimuli reverses at slow presentation rates. Explanations for these differences in recall for single‐ and multiple‐talker word lists have suggested that processing multiple‐talker stimuli either increases rehearsal demands during transfer to long‐term memory or increases difficulty of perceptual encoding. The present experiment examines these competing explanations for differential serial recall in multiple‐ versus single‐talker word lists using voices that subjects have been trained to identify. As listeners become familiar with a talker's voice, it is possible either that mandatory rehearsal will encompass richer recall cues, including talker‐specific information, or that perceptual normalization will become more automatic. Results provide support for the proposal that rehearsal of enriched cues leads to the differences in recall for multiple‐ and single‐talker lists. Using familiar voices increases serial recall for multiple‐talker lists over single‐talker lists in the primacy portion of the recall curve, but does not appear to facilitate perceptual encoding. [Work supported by NIH Grant DC‐0111‐14 to Indiana Univ.]