India is the land of spices and is the largest producer, consumer and the exporter of spices in the world. Spices are an important component of Indian Agricultural Exports earning valuable foreign exchange and are the source of livelihood for millions of small and marginal farmers across different states of the country. Modeling of agricultural exports in general and spices exports in particular is important in the contest of spices exports being a priority area for Indian policy makers. Time series modeling of agricultural commodity exports is an active area of research in recent times. Generally Box Jenkins approach (ARIMA) is the referred technique for this purpose. When data exhibits volatility clustering, ARCH/GARCH models are used. When the data does not support linearity... assumptions neutral network models are used. However, real world time series data is believed to be a combination of linear and non-linear patterns. In this context, Hybrid models which are a combination of AR models and Artificial Neural Networks are providing more accurate forecasts. The present study, using secondary data for the period from 1960–61 to 2017–18 applies three hybrid models for forecasting Indian spices exports both in terms of volume and prices. Based on the RMSE each model is evaluated and finally model with least RMSE was selected for forecasting both volume and unit prices of total spices export for the coming 10 years (2018–19 to 2027–28). Analysis of data was done with the help of open source software R. Results from the study show that, Hybrid model consisting of ARIMA, Exponential Smoothing and Tbats Model with unequal weights was found to be the best model on the basis of RMSE for forecasting Indian spices exports. Thus, for both forecasting and policy formulation the hybrid model is recommended.