In voice spectrogram analysis, an expert operator identifies an unknown voice by comparing graphic displays produced by the spectrograph along with speech samples. The existence of variations within one person's speaking voice has cast doubt on the accuracy of spectrogram analysis, even though research studies have indicated that variations within one speaker's patterns are less than those among different speakers. Courts have accepted evidence based on anatomical uniqueness, such as fingerprints and blood samples, but have been reluctant to admit the validity of voice analysis. This can be attributed to the lack of knowledge about the effects of surgery or deliberate changes in speech on the spectrogram's reliability and a 6 percent error rate in false identifications during open trials reported by a major 1972 research project. These unresolved issues, however, are not significant enough to exclude all spectrographic evidence. Many courts have continued to rely on the Frye versus United States decision, concerning the inadmissibility of polygraphs, as evidence to deny the admission of spectrogram evidence. This approach is inappropriate because spectrogram evidence is inherently less dangerous and less conclusive than polygraph results. Furthermore, the Frye doctrine may be misconstrued by using a broad definition of a scientific field rather than narrowing it to include only those experts who are familiar with a specific technique. Under this interpretation, a good case for admissibility of voice spectrograms can be made because most experts on the subject support the technique's reliability. The Federal Rules of Evidence constitute a preferable alternative to Frye since the rules would allow more spectrogram evidence and yet exclude misleading and irrelevant data. Once the basic reliability of a new scientific technique is known, these guidelines permit the court to weigh the prejudicial effect the evidence may have against its probative value in the case. The article contains 80 footnotes.