This article explores the relation between serial correlation and volatility for several different stock return series at daily and weekly frequencies. It is found that serial correlations are changing over time and are related to stock return volatility. An extension to the GARCH model is proposed and estimated, revealing parameters consistent with other findings in this article.