In this paper we test for the presence of rational bubbles in the NASDAQ stock market index over the period 1994:06–2003:11 by means of a methodology based on fractional processes. The results suggest that the existence of bubbles depends on the sampling frequency used in the analysis. We cannot reject the unit root hypothesis when using monthly data on price–dividend ratios, which according to the present value model suggests the existence of rational bubbles. However, we reject this hypothesis in favor of fractional alternatives when using daily and weekly data. This might be explained by the temporal aggregation and/or the sample sizes used in the application.