The dissertation investigates the starting transient behaviors of metal halide lamps driven by constant currents and constant powers, respectively. Based on the investigation results, three starting scenarios are proposed for shortening the starting time, and an identification strategy is figured out for designing an electronic ballast being capable of driving three small-wattage lamps rated at different powers. A laboratory electronic ballast is designed to drive small-wattage metal halide lamps with a programmable low-frequency square-wave current. Experiments are conducted to examine the effects of the starting current on variations of the light output as well as the lamp voltage and power. From the effects of the applied current on the generated luminance, three starting scenarios are attempted to accelerate the starting transient stage. Experimental evidence shows that the starting time can be effectively shortened by increasing the lamp current during glow-to-arc and warm-up stages. A short interval of over-power operation during the warm-up stage enables the lamp to further enhance the producing of luminance quickly, and hence greatly reduce the starting transient period. According to the starting transient characteristics of metal halide lamps, an identification strategy is figured out to recognize three small-wattage metal halide lamps rated at powers of 20-W, 35-W and 70-W from three world-wide prominent brands, GE, OSRAM and PHILIPS. An electronic ballast is designed to drive the metal halide lamps with the multi-stage constant-power starting scenario. Experimental results evidence that the electronic ballast with the proposed identification strategy can recognize three lamps¡¦ rated powers correctly during the starting transition, and drive the lamp to its rated power before entering the steady-state.