We used DNA fingerprinting by the arbitrarily primed polymerase chain reaction (AP-PCR) technique for an epidemiological investigation of 23 Pseudomonas cepacia isolates obtained from 11 cystic fibrosis (CF) patients attending our CF center. This approach was compared with ribotyping, pulsed-field gel electrophoresis (PFGE), and conventional phenotypic typing. AP-PCR and ribotyping were identical in resolving power, since the two methods generated four different profiles and identified the same group of strains. Six patients on the one hand and four on the other harbored strains of the same genotype, thus raising the possibility of either patient-to-patient transmission or acquisition from a common hospital environmental source. PFGE results were in good agreement with those of the other two methods, but PFGE seems more discriminative since it generated a fifth profile for a single strain in a group of four. Our results show in vivo stability for the three methods during a period extending from 3 to 41 months. These genotypic techniques are particularly promising for clinical laboratories to help to clarify the epidemiology of P. cepacia in CF patients. The AP-PCR method constitutes an easier alternative to the well-established ribotyping method. AP-PCR provides the quickest results with minimal technical complexity. However, our results suggest that it is less discriminative than the labor-intensive PFGE method.