The Asian financial crisis, which has hit most countries in East Asia since 1997, has drawn the attention of many researchers dealing with the financial and economic sectors. Under these severe financial crises, their currencies fell more than 50 percent and the stock market plunged about 30 to 50 percent. The purpose of this study was to investigate volatility transition between the returns of stock markets of Malaysia, Singapore and Thailand in ASEAN's major economies, using monthly data from 1990 to 2020. The results of this study showed that there were no autocorrelation or serial correlation in the ADF test because there was no unit root for each level variable. In addition, we found no evidence of a cointegration relation(vector) between the three stock return series, indicating there was no long-term equilibrium relationship in the Johansen cointegration test. Thus, instead of modeling all stock returns using a linear vector autoregression(VAR) model, this paper used the multivariate Markov switching vector autoregression model(MS-VAR) with two regimes and 1 lag, including that the stock returns were regime dependent and whose mean and variance both had regime changes. And the study found that it was successful in capturing satisfactory timing of several financial crises that occurred in three Asian countries in connection with the 1990(oil crisis), 1997(foreign currency crisis), 2007(U.S. financial crisis), 2008(Global financial crisis), and 2020(Corona 19 pandemic crisis). Consequently, we found that the MS-VAR model was successful in identifying common regime changes in volatilities of the three stock return series, and showing remarkable evidence of significant interrelation between volatilities of the stock returns in Asian countries.