We propose a novel approach to forecasting core inflation in India, whose average contribution to headline inflation has been about 55 percent since January 2016. Our approach involves using the dis-aggregated components of core inflation, as well as the construction of a demand index using high frequency (HF) indicators. We find that individually forecasting and then aggregating core CPI components improves the short-term forecasting accuracy of core inflation. However, forecasting aggregate core inflation directly is more effective for longer horizons. We estimate a demand index using high frequency indicators. We find that the inclusion of the demand index and other co-variates enhances forecasting efficacy by capturing demand-side factors specific to the Indian economy. We also find that an accurate specification of the dis-aggregate components model contributes to maximizing prediction accuracy.