This study applies fractional integration techniques (parametric and semi-parametric) to examine the time series behavior of US retail gasoline prices using weekly data from January 2, 1995, to May 22, 2017. The results based on both parametric and semi-parametric methods provide conflicting evidence. Using parametric methods, the orders of integration are significantly greater than 1; however, with the semi-parametric approach, there is some evidence of mean reversion (d < 1). These conflicting results could be due to the existence of structural breaks. Indeed, endogenous structural break tests indicate breaks in 2005 and 2010. The estimation of the orders of integration within each of the subsamples reveals the absence of mean reverting behavior in retail gasoline prices.