A comparison was made of three approaches for the detection of Listeria monocytogenes in naturally contaminated soft cheese and semi-soft cheese after enrichment. Enrichment broths were tested by plating them onto different selective agars, by "Gen Probe" DNA hybridization and by the polymerase chain reaction (PCR). Based on two-step enrichment, all three approaches showed high specificities (90% or more) in detecting L. monocytogenes. In contrast, the sensitivity of the Gen-Probe test was low (33% or less), whereas high sensitivities were obtained with selective plating and PCR (83% or more). Based on one-step enrichment, specificities again were high for selective plating and PCR assay (100%), whereas for the Gen-Probe assay the specificity was lower (88% or more). The best sensitivities were observed with selective plating (67%) and PCR (75%). In terms of sensitivity, specificity and analysis time, PCR applied to a two-step enrichment was the most powerful assay for detecting L. monocytogenes in soft and semi-soft cheese.