A PCR-detection method for M. paratuberculosis in milk has been evaluated. Raw milk, aseptically drawn, was spiked with decimal dilutions of M. paratuberculosis cells. Enumeration was performed by: epifluoresence microscopy (DEFT), solid phase cytometry (ChemScan; Chemunex) after viability labelling with a fluorogenic substrate for esterase and by BactoScan 8000 (Foss Electric). For PCR detection, the milk was first chemically or enzymatically destroyed and the bacterial cells were collected by centrifugation according to Herman et al. (Appl. Environ. Microbiol. 61:817-819, 1995) and Rijpens et al. (Appl. Environ. Microbiol., 62:1683-1688, 1996), respectively. M. paratuberculosis was specifically detected by a nested PCR according to Millar et al. (Appl. Environ. Microbiol. 62:3446-3452, 1996). The enzymatic destruction of the milk resulted in a 10 times higher detection sensitivity compared to the chemical destruction. This is probably due to the affinity of mycobacteria for milk fat, which is more efficiently removed in the enzymatic method. About the same results were obtained with the DEFT and ChemScan methods, while the BactoScan showed a considerable variation because the intensity of the pulses approached the discrimination level of the apparatus. This is probably due to the specific cell wall composition of mycobacteria. In one experiment there was a 5 times underestimation, in a second experiment the underestimation increased with the concentration from 6.6 to 20. The PCR enzymatic based detection reached an average sensitivity of 30 cfu/ml. These results indicate that M. paratuberculosis cells could be efficiently detected in raw milk by the enzymatic based PCR method. Although no specific detection was achieved with the ChemScan using the above viability labelling, it could be promising for the automatic and accurate enumeration of M. paratuberculosis in milk.