A voiceprint of a calling party, and identifying information, are obtained during a call. One or more voiceprint samples stored in association with the identifying information are selected, and the calling party's identity is verified based on a comparison of the voiceprint to at least one of the selected voiceprint samples. For example, the voiceprint samples may be derived from a verbal utterance by the calling party before the call, mixed with different background noises likely corrupting the calling party's utterance during the call. If the comparison fails to determine the calling party's identity, an attempt to verify the calling party's identity is made based on additional information received during the call. In another example, if the calling party's identity is verified based on the additional information, the voiceprint may be stored in association with the identifying information as a voiceprint sample for use in future verification of the calling party's identity.