In this paper, we assess the impact of regime shifts on the long memory properties of the Indian exchange rates. We make use of Sanso, Arago and Carrion (2004) Iterated Cumulative Sum of Squares (hereafter referred as AIT-ICSS) algorithm to detect the points of structural breaks in volatility series. Our findings indicate that incorporating the impact of sudden changes in volatility in the model indeed reduces the magnitude of long memory parameter. In the case of INR/JPY, we observe a shift in characteristics from long memory to mean reversion when the impact of regime shifts is included in the volatility model. Our findings also highlight that incorporating the impact of regime shifts in the model also improves the volatility forecast accuracy. Moreover, we implement a trading strategy based on risk-averse investor and find that the volatility forecasts based on the model which incorporate the impact of structural breaks provide substantial gains in return in comparison to volatility models with no structural breaks. These findings have important policy implications for financial market participants, investors and policy makers.