Voice identification evidence, identifying an offender by the sound of their voice, is sometimes the only means of identifying someone who has committed a crime. Auditory memory is, however, associated with poorer performance than visual memory, and is subject to distinctive sources of
unreliability. Consequently, it is important for investigating authorities to adopt appropriate strategies when dealing with voice identification, particularly when the identification involves anvoice previously unknown to the witness. Appropriate voice identification parades conducted by
police can offer an otherwise unavailable means of identifying the offender. This article suggests some ‘best practice’ techniques for voice identification parades and then, using reported Australian criminal cases as case studies, evaluates voice identification parade procedures used by police. Overall, we argue that the case studies reveal practices that are inconsistent with current scientific understandings about auditory memory and voice identifications, and that courts are insufficiently attending to the problems associated with this evidence.