The present study investigated the Accumulated Oxygen Deficit (AOD) method in front crawl swimming with the aims to assess the robustness of the oxygen uptake/swimming velocity regression line and to quantify the precision of the AOD. Twenty-nine male swimmers performed two swimming tests in different days, with a 24 h recovery between tests: a graded test and an all-out test. The all-out test was performed either in 100 m (n=11), 200 m (n=13) or 400 m (n=5). Through all testing expired gases were collected breath by breath and analysed with a K4b2 Gas Analyser (Cosmed, Rome, Italy) connected to an AquaTrainer Valve (Cosmed, Rome, Italy). The error of oxygen uptake/swimming velocity regression lines was approximately 4-5 ml x kg (-1) x min (-1)) and the regressions allowed an extrapolation of the energy cost to higher intensities with a standard error of prediction of approximately 3-4 ml x kg (-1) x min (-1). However, the data variability was considerable (95% confidence intervals of the linear extrapolation larger than 13 ml x kg (-1) x min (-1)). The AOD imprecision varied among the three distance events from approximately 3-13 ml x kg (-1). These absolute errors are small, considering the time that subjects took to cover the three distances, but relative to the AOD values that were estimated they can be considered high, especially in the 400 m bout.