For faster moving non-trended products, it is well established that Single Exponential Smoothing is more accurate than Simple Moving Averages for short-term forecasting (Makridakis et al (1982)). However, much of the benefit that derives from SES can be obtained by the simpler expedient of using a combination of Simple Moving Averages (Johnston et al (1999), Boylan and Johnston (2003)). In an intermittent demand setting, Croston (1972) proposed using a ratio of size and interval estimates, both based on exponential smoothing. This method has been used extensively in practice. Recently, Syntetos and Boylan (2001) showed that this method is biased and proposed an alternative estimator, using a correction factor. In this paper, a new estimator is proposed, based on a ratio of Simple Moving Averages of Size and Interval, allowing different lengths of averaging to be used on the numerator and denominator if necessary. A correction factor is presented, to ensure that the method is approximately unbiased. The performance of the two estimators is assessed using empirical data from the perspective of forecasting accuracy, approaches, is outlined.