With the need to improve the operation and control of existing generating units and to provide accurate information for the design of new units, better transfer function models and estimates of system parameters are required. This involves the identification of individual generator and voltage regulator parameters as well as system modes of oscillation and damping. The problem is to experimentally determine this information, simply, quickly and accurately. For this reason, system identification techniques are of interest to power engineers. In the paper, an identification technique involving pseudo-random ternary noise injection and cross correlation is assessed and applied to power systems. The theoretical basis, advantages and limitations of this method are examined and it is shown that the entire process of signal generation, injection, response measurement and crosscorrelation can be performed in real time using a dedicated microprocessor system. The identification technique is tested on a simple microalternator connected to a noisy power system through a transmission line. It is shown that under normal operating conditions a highly accurate system impulse response can be obtained and that the parameters of the machine and its voltage regulator can be estimated from this result. The technique is also applied to an open-circuited microalternator with a more complex voltage regulator and estimates of the regulator parameters are found with good accuracy.