Implied expected returns are the expected returns for which a supposedly mean-variance efficient portfolio is effectively efficient given a covariance matrix. We analyze the statistical properties of monthly implied expected return estimates and study their sensitivity to the choice of a mean-variance efficient portfolio proxy. Over the period January 1984 to December 2012 and for the universe of S&P 100 stocks we find that the largest gains are in terms of stability of the return forecasts. The use of a maximum diversification or equal-risk-contribution portfolio as proxy reduces significantly the cross-section and time series dispersion in the implied expected return forecasts and leads to a small improvement in forecast precision, compared to using a market capitalization, fundamental value or equal weighting scheme. For all proxies considered, the implied expected return estimates outperform the time series model based forecasts in terms of stability and forecast precision.