capillary electrophoresis (CE) is characterized. A 300 mum diameter Pt working electrode was used to directly couple with a 75 mum inner diameter separation capillary without an electric field decoupler. The hydrodynamic cyclic voltammogram (CV) of Ru(bpy)(3)(2+) showed that electrophoretic current did not affect the ECL reaction. The presence of high-voltage (HV) field only resulted in the shift of the ECL detection potential. The distance of capillary to electrode was an important parameter for optimizing detection performance as it determined the characteristics of mass transport toward the electrode and the actual concentration of Ru(bpy)(3)(2+) in the detection region. The optimum distance of capillary to electrode was decided by the inner diameter of the capillary, too. For a 75 mum capillary, the working electrode should be placed away from the capillary outlet at a distance within the range of 20-260 mum. The effects of pH value of ECL solution and molecular structure of analytes on peak height and theoretical plate numbers were discussed. Using the 75 mum capillary, under the optimum conditions, the method provided a linear range for tripropylamine (TPA) between 1 x 10(-10) and 1 X 10(-5) mol/L with correlation coefficient of 0.998. The detection limit (signal-to-noise ratio S/N = 3) was 5.0 x 10(-11) mol/L. The relative standard deviation in peak height for eight consecutive injections was 5.6%. By this new technique lidocaine spiked in a urine sample was determined. The method exhibited the linear range for lidocaine from 5.0 x 10(-8) to 1.0 X 10(-5) mol/L with correlation efficient of 0.998. The limit of detection (S/N = 3) was 2.0 x 10(-1) mol/L.