O The present study aims to explore predictability in temporal dynamics regarding linear models of the implied volatility surfaces estimated for Brazilian stocks options. Previous results, by usual approach of fitting linear models linking implied volatility to time to maturity and moneyness, available for each cross-section of option contracts at a point in time, as in Dumas, Fleming and Whaley (1998), suggest that estimated parameters of such models are highly unstable over time. Therefore, this approach isn t capable of replicating various IVS s shapes, contrary to the empirical evidence of implied volatility varying with options strike price and date of expiration. Based on these evidences and in Heston and Nandi (2000), that exploit the information on path-dependency in volatility contained in the spot S&P 500 index, Gonçalves e Guidolim (2006) proposed a two-stage approach to modeling and forecasting the S&P 500 index options IVS. In the second-stage they model the dynamics of the cross-sectional first-stage coefficients by means of vector autoregression models. The contribution of this work is to apply the proposed model to the reality of the Brazilian stock options, incipient in terms of liquidity and trading horizon dimensions when compared to the U.S. market, adapting criterians to validate its applicability in this context in statistical and economical sense. The results demonstrate the superiority of this approach over comparable literature, but not the ability to generate abnormal profits in the presence of transaction costs in excess of the benchmark of the risk-free rate. This indicates adaptation to the market efficiency hypothesis.