The API 20S (Analytab Products, Plainview, NY), the GPI card (Vitek Systems, St. Louis, MO) and the RapSTR system (Innovative Diagnostics, Atlanta, GA) were compared with conventional biochemicals for the identification of viridans streptococci. One hundred nine clinical isolates were tested that included the following species: intermedius (38) sanguis II (20), bovis (variant) (14), mitis (14), salivarius (11), sanguis I (6), constellatus (3), mutans (2), and uberis (1). With initial testing, a correct species call was made with 72% of the isolates with the GPI card, 62% with the RapSTR, and 50% with the API 20S. Identifications of viridans streptococci group or those that needed additional biochemicals for species identification occurred with 28% of isolates with the API 20S, 8% with the RapSTR, and 9% with the GPI card. Incorrect identifications occurred with 6% of the isolates tested by the GPI card, 20% with the API 20S, and 30% with the RapSTR. Most discrepancies with the RapSTR were with 66% of the intermedius isolates, whereas most, 55%, of misidentifications with the API 20S were with sanguis II isolates. No identifications were made with 2% and 13% of isolates with the API 20S and GPI, respectively.