In Multiple Input Multiple Output (MIMO) systems, predicting performance prior to data transmission is needed to appropriately select transmitter parameters (e.g. power, code rate). The prediction must be both accurate and easy to compute, given limited resources. An algorithm taking too much time to run or excessive circuitry to implement cannot be deployed. This paper presents an algorithm to predict the average bit error rate (BER) and packet error rate (PER) of a MIMO system, so that prediction is not only accurate, but is also of low computational complexity. Since the algorithm makes no a-priori assumptions about the channel, it is applicable to a wide variety of random channels. In our results, we show a good match between the predicted BER and PER, and the BER, PER from simulated data transmission.