The capillary electrochromatographic (CEC) analysis of basic compounds on octadecyl-silica stationary phases (Hypersil ODS and Spherisorb ODS I) was studied. A basic drug (fluvoxamine) and one of its possible impurities were used as test compounds. With an eluent of acetonitrile-phosphate buffer (pH 7.0), the compounds could be baseline-separated; however, broad and tailing peaks were obtained. To minimise detrimental interactions with residual silanol groups, the pH of the mobile phase was lowered to 2.5, but the plate numbers were still quite low (<2.6x10(4) plates/m). Addition of a masking agent (hexylamine or triethylamine) to the mobile phase resulted in much better peak efficiencies (ca. 1x10(5) plates/m). Therefore, the influence of the amine concentration and pH of the mobile phase on the CEC performance (peak width, peak tailing, electroosmotic flow, selectivity) was investigated in detail. Highest efficiencies (2.8x10(5) plates/m) could be obtained with the Spherisorb column, while the Hypersil column offered a better selectivity. Furthermore, the results show that the residual silanol groups are (at least partly) responsible for the separation of the basic compounds and that the amount of injected sample has an unusually large effect on the peak efficiency. The usefulness of the system for impurity profiling was demonstrated with a mixture containing fluvoxamine and its stereoisomer (a possible impurity) at the 0.1% level. The general effectiveness of amine additives in CEC was illustrated by the separation of a mixture of five structurally different basic drugs yielding plate numbers in the 1x10(5)-3x10(5) plates/m range. Comparison with capillary electrophoretic analysis revealed a unique selectivity of the CEC system which is based on both electrophoretic mobility and chromatographic partitioning.