This paper develops a robust test for long-term memory in stock market prices, extending Mandelbrot's "range over standard deviation" (R/S) statistic. The modified R/S statistic is designed to account for short-term autocorrelations, which can bias the classical R/S statistic. The test is applied to daily, weekly, monthly, and annual stock return indexes over various time periods. Contrary to previous findings, the data do not show evidence of long-term memory when short-term autocorrelations are considered. Monte Carlo experiments indicate that the modified R/S test has power against specific models of long-term memory, suggesting that conventional stochastic models with short-term dependence may adequately capture the time series behavior of stock returns. The results reinforce the idea that the long-run predictability of stock returns may be due to short-term models rather than long-term memory.This paper develops a robust test for long-term memory in stock market prices, extending Mandelbrot's "range over standard deviation" (R/S) statistic. The modified R/S statistic is designed to account for short-term autocorrelations, which can bias the classical R/S statistic. The test is applied to daily, weekly, monthly, and annual stock return indexes over various time periods. Contrary to previous findings, the data do not show evidence of long-term memory when short-term autocorrelations are considered. Monte Carlo experiments indicate that the modified R/S test has power against specific models of long-term memory, suggesting that conventional stochastic models with short-term dependence may adequately capture the time series behavior of stock returns. The results reinforce the idea that the long-run predictability of stock returns may be due to short-term models rather than long-term memory.