In order to make a fast identification of M. paratuberculosis in culture, a PCR based on the IS900 gene was introduced and evaluated on suspected colonies found in the routine diagnostic work. Except for one case, there has been agreement between the PCR identification and the conventional methods based on growth characteristics, acid-fastness and mycobactin dependence. The isolates of acid-fast bacilli obtained from routine cultures were compared for mycobactin dependence and IS900 PCR reaction with the p36/p11 primers. In one case, a single colony suspected to be M. paratuberculosis was detected from a faecal sample of a healthy dairy cow in Western Sweden. The isolate had acid-fast bacilli, a little longer than usual for M. paratuberculosis and some bacilli were bent. It tested positive with the PCR for IS900, but at subculture it turned out not to be mycobactin dependent. Genetic probe against Mycobacterium avium complex (Accuprobe, GenProbe, San Diego, USA) did not hybridise with this strain as should have been expected for M. paratuberculosis. By sequencing the 16S rRNA gene, this isolate was shown to be unrelated to M. paratuberculosis. Instead, it grouped with M. cookii and Mycobacterium sp. strain IMVS B76676. The similarity values for strain 2333 to M. cookii and Mycobacterium sp. strain IMVS B76676 were 98.3% and 98.8%, respectively. The similarity value to M. paratuberculosis was only 96.4%. Five other PCR systems targeting IS900 were used to test this isolate and they were all positive. Restriction endonuclease analysis of the IS900 PCR products according to Cousins et al. (1999) did not reveal any differences as compared to the IS900 PCR products from M. paratuberculosis. Our results clearly demonstrate that a positive IS900 PCR alone does not prove the presence of M. paratuberculosis. Further confirmation has to be made, for instance by isolation and fenotypic identification or by identification with an alternative PCR.