THE CONCEPT OF RISK which has received so much attention in academic circles for more than two decades has now begun to have an impact on practicing portfolio managers. The impetus for this has been gradual acceptance of the efficient-market hypotheses because of the inability of most portfolio managers to outperform the popular stock market indices in recent years and the deveopment of modern capital market theory. The purpose of this paper will be to examine the predictability and stationarity of real portfolio risk levels utilizing mutual fund portfolios that have not adopted an indexing concept but that do have explicit investment objectives that should provide a guide as to the future risk exposure of the portfolio.' The implications of a stationary risk level would be favorable to investors who could then purchase mutual fund shares in portfolios having a risk exposure equal to their own risk preferences and/or lever the risk of the portfolio either by borrowing additional funds or investing in a riskless asset. Furthermore, in evaluating portfolio performance, an implicit assumption is that the riskiness of the portfolio has not changed over the period under consideration.2 This study investigates whether mutual fund risk levels are stationary and therefore predictable enough to be used by investors in making investment decisions and measuring ex post risk-adjusted performance. Some absolute standards with which to compare the stationarity or predictability of risk have fortunately been established by others, notably Blume [1] and Levy [8]. These studies and others will be reviewed in the first section, the sample data is explained in the second section and the methodology and empirical results provided in the third section.