This paper relates variation in stock market volatility to regime shifts in stock market returns. We apply a Markov switching model to market returns and examine the variation in volatility in different return regimes. We find that stock returns are best characterized by a model containing six regimes with significantly different volatility across the regimes. Volatility is higher when returns are either above or below the normal regime-- the further returns deviate from the normal regime, the higher the volatility. Furthermore, volatility is higher in negative return regimes than in positive return regimes. These observations lead us to conclude that return and volatility are related nonlinearly and that the relationship is asymmetric.