Flow injection (FI) sample introduction was coupled with a recently developed atomic emission source, the microwave plasma torch (MPTJ, operating at 200 W with Ar as the plasma gas. The use of FI in conjunction with an ultrasonic nebulizer was compared with direct nebulization; parameters taken into consideration include carrier solution flow-rate, aerosol carrier gas flow-rate, sample volume, dispersion coefficients and.the effect of two easily ionized concomitant elements. Detection limits and precision of the FI technique were assessed for four sample-loop volumes: 75, 200, 250 and 500 ~1. Flow injection was found to offer several advantages over continuous sample introduction: rapid sample throughput and a reduction of memory effects without a loss in sensitivity or precision. Furthermore, by appropriate choice of sample dispersion, a significant reduction of the Na and K interferences can be achieved without a substantial deterioration in sensitivity. An FI system incorporating a microcolumn of bonded silica with octadecyl functional groups was used to determine Cu in synthetic sea water. The sample preconcentration time was one minute at a sample solution flow-rate of 1.0 ml nun-'; pure ethanol was used as the eluent. The detection capability in this preconcentration mode was approximately 25 times better than that offered by a comparable continuous nebulization system. The detection limit for Cu was 0.16 ng ml-', the relative standard deviation at 500 ng ml-' was about 3.2%, and recovery rates were in the 94-97% range.