In many countries, coagulase-negative staphylococci (CNS) are currently the most common cause of intramammary infection in lactating cows. In order to elucidate the importance of various CNS species in udder health and milk quality, further research conducted on the species level is required. Phenotypic identification of CNS species appears to be unreliable and more accurate and reproducible genotypic methods are needed. In the current study, use of amplified fragment length polymorphism (AFLP) genotyping was validated for species identification of bovine associated CNS. An initial reference library was generated with AFLP fingerprints of 52 different CNS type and reference strains. Next, 247 bovine CNS field isolates with known species identity were analyzed. These field isolates had been previously identified by gene sequencing and were randomly divided into two subsets, i.e. a training set and a validation set. The training set was identified against the initial reference library containing only type and reference strains, which resulted in a typeability of 80.5%. Accuracy of the AFLP identifications, being the correspondence with gene sequencing results, was 95.0%. Fingerprints of the training set were then added to the initial library and identification of the validation set was done by means of this extended library. By adding bovine CNS to the library, performance of the AFLP identification method improved considerably. Final typeability and accuracy were 98.4% and 99.2%, respectively. Numerical analysis of AFLP fingerprints proves to be an accurate genotypic method for identification of CNS from bovine origin. The constructed AFLP library provides a useful identification tool for field studies on the subject of CNS.