We present a new procedure for detecting additive outliers in GARCH(1,1) models. The outlier candidate is the observation with the largest standardized residual. First, a likelihood-ratio based tests determines the presence of an outlier. Next, a second LR test determines the type of outlier (volatility or level). This procedure is shown to be approximately independent from the GARCH parameters, with a null distribution that can be easily approximated. We apply the method to returns of the Dow Jones index, using monthly, weekly, and daily data. The procedure is extended to cover GARCH models with Student-t distributed errors.