Over 30 sea surface microlayer (SML) samples from two contrasting sites in the North Western Mediterranean -- Barcelona (Spain) and Banyuls-sur-Mer (France) -- were collected using three different sampling devices, namely, glass plate, metal screen (MS) and a surface slick sampler (SS), and compared with the corresponding underlying water (16 samples). The distributions of 41 polychlorinated biphenyl congeners (PCBs) were determined in the different phases: particulate (1.17-10.8 SigmaPCB ng L(-1)), truly dissolved (0.080-16.7 SigmaPCB ng L(-1)) and colloidal matter (1.17-43.0 SigmaPCB ng L(-1)), being the last two estimated from the analysis of the apparently dissolved phase. Concentrations of PCBs in the SML were higher than those in the underlying water (ULW), giving rise to enrichment factors (EF=[C](SML)/[C](ULW)) up to first-order of magnitude. The ANOVA statistical approach was used to assess differences of bulk data (e.g. dissolved organic carbon, DOC; particulate organic carbon, POC; suspended particulate matter, SPM) among sampling devices, whilst p-tailed t paired tests were used in order to compare the enrichments obtained for each sampling date. In this respect, no significantly different enrichment factors were found among sampling devices (p < 0.05), although the surface SS showed lower enrichments, probably due to the dilution of the SML with the ULW during sampling. The MS seemed to be the most suitable device for the determination of PCBs in the SML in terms of sampling efficiency under a variety of meteorological conditions.