Mathematical descriptions for extraction recovery and enrichment were applied for liquid-phase microextraction (LPME) and comparison with conventional two- and three-phase liquid-liquid extraction techniques (LLE) was made. The LPME theoretical calculations were verified by experimental determination of actual partition coefficients and by data obtained with LPME in a robust hollow fibre formate. With hollow fibre LPME operated in the two-phase mode, analytes were extracted from 1 to 4 ml aqueous samples into 25-50 μl of an organic solvent present in the pores and in the lumen of the porous hollow fibres. Compared with conventional two-phase LLE, two-phase LPME provided substantially higher enrichments for compounds with relatively large partition coefficients (K_org/d>500). In contrast, because of the large volume of organic solvent relative to the sample volume, LLE provided high recovery and moderate enrichment even for compounds with relatively low partition coefficients (K_org/d>5). Thus, two-phase LPME may be used for substantially enhanced extraction selectivity and enrichment of relatively hydrophobic analytes as compared with LLE whereas conventional two-phase LLE is superior for more hydrophilic analytes. Similar results were found for three-phase LPME where analytes where extracted from 1 to 4 ml aqueous samples through approximately 20 μl organic solvent immobilized within the pores of the hollow fibre and into 25 μl of an aqueous acceptor solution inside the lumen of the hollow fibre. The fundamental differences of LPME and LLE were further demonstrated with practical experiments on extraction of the basic drugs promethazine, methadone, and haloperidol from human plasma and urine.