Exchange rates are important financial indicator that is receiving attention from researchers globally. This study uses daily data over the period January, 1999 to February, 2014 consisting of 3950 observations. The study aimed at determining the volatility spillover of the world's major currencies against the U.S. Dollar simultaneously using Multivariate GARCH models. And to observe some stylized facts/common features of good volatility modeling on financial time series. To achieve the objectives, we employed three volatility models: DVECH, BEKK and CCC and the results indicated that the skewness is greater than zero, implying that the distribution is positively skewed. The kurtosis is also greater the kurtosis of a normal distribution: the distribution of the exchange rates return series is leptokurtic. We concluded that the BEKK model proved to be the best model because it has maximum log likelihood and least number of parameters. We also found that the volatility of the Nigerian Naira is more stable than the rest of the currencies in Africa; this implies that the investors are at low risk and have chances of making more profits in Nigeria than in any other African country.