This research paper aims to examine the predictability of the Spanish Stock Market returns. Earlier studies suggest that stock market returns in developed countries can be predicted with a noise term but this study has specifically covered two time horizons; one pre-crisis period and the other one current crisis period to evaluate the stock market returns predictability. Since mean returns cannot prove all the time to be efficient predictor, variance of such returns do, hence various autoregressive models have been used to test the existence of persisting volatility in the Spanish Stock Market. The empirical results show that higher order of autoregressive models such as ARCH(5) and GARCH(2, 2) can be used to predict future risk in Spanish Stock Market both in pre-crisis and current crisis period. The paper also reveals that there is a positive correlation between Spanish Stock Market returns and the conditional standard deviations as produced by ARCH(5) and GARCH(2, 2), implying that the models have some success in predicting future risk on Spanish Stock Market. The predictability of stock market returns during crisis period is not found to be affected contrary though the degree of predictability may be. © 2021 by World Scientific Publishing Co. Pte. Ltd.