The previous chapter provided the empirical evidence of volatility modelling of exchange rate markets in Malaysia in a univariate framework. This chapter looks at the analysis of exchange rate volatility in a multivariate framework. From the finance point of view and whenever portfolios of assets are involved, the univariate approach (as discussed in Chapter 4) has some severe shortcomings, as one has to deal with the replicating portfolio to get back to the univariate case and use the various risk measurement models like VaR models. In addition to the flexibility of the multivariate approach, these models account for the increasing correlation between returns during periods of high volatility. Section “Introduction” introduces the rationale for extending the analysis of exchange rate volatility modelling to a multivariate framework. Section “Framework and estimation process” analyses the literature and the concept of exchange rate volatility modelling in a multivariate framework. The empirical results are discussed in section “Estimation processes” while section “Diagnostic testing” discusses the implications of the findings in the multivariate cases. Finally, section “Empirical results” concludes this chapter. Applied researchers and investors who are doing research for developing markets may want to consider one or more of the four MGARCH models discussed in this chapter to simulate the conditional correlations for the purpose of portfolio analysis. The conclusion is that to model the conditional correlations of the currency pair of a small open economy such as Malaysia it is reasonable to use MGARCH models of DCC models and BEKK models. One however cannot use these MGARCH models for meaningful risk measurements and will need to fall back to univariate GARCH models.