A discrete time domain-based system identification method for PWM DC-DC converters is proposed. Accurate information on the system's open-loop response is essential in the design of the system controller in order to obtain the desired closed-loop response. This is especially true in switch-mode converters where component uncertainty exists. It is conjectured in this study that an identification method that is based on time-domain signals will be relatively simple to realise with a digital processor. The method that is proposed is capable of successfully reconstructing the system model by an arbitrary excitation at the command input. In this study, a step perturbation was employed, which is simple to apply and leads to an intuitive interpretation of the output response. The effects of switching and quantisation noise have been overcome by choosing the sampling interval after the switching oscillations have decayed and by averaging the responses of synchronously perturbed sequences. The proposed method has been evaluated on Buck and Boost converters. The method was verified in two phases: Off-line – data acquisition procedure was implemented on a TMS320F2407-DSP and the identification calculations were carried out on a PC. On-line – The identification procedure (data acquisition plus fitting algorithm) was programmed on a TMS320F2808-DSP.