<p>Soil detachment is a key parameter to describe the hydrosedimentological processes in watersheds. Few studies have assessed the applicability of the existing equations for calculating soil detachment in specific environments such as unpaved roads with compacted soil. The detachment efficiency coefficient of soil is a dimensionless parameter that uses soil cohesion to determine the amount of soil disaggregated by the shear force of runoff. The aim of this study was to compare monitored with estimated data series of runoff (<i>Q</i>) and sediment production (<i>Q</i>_<i>s</i>) calculated by the main three methodologies used for estimation of soil detachment efficiency coefficient (Rauws and Govers 1988; Morgan et al. 1998; and Morgan 2001), available in the openLISEM rainfall-runoff-erosion model, for a section of an unpaved rural road, bounded by water bars, in a <i>Pinus taeda</i> L. forest in Southern Brazil, over contrasting rainfall events monitored on summer of 2013. The simulation results were also compared with each other based on runoff erosion (<i>E</i>_<i>r</i>), splash erosion (<i>E</i>_<i>s</i>), deposition (<i>D</i>), average soil loss (<i>Qa</i>_<i>s</i>) and total soil loss (<i>Qt</i>_<i>s</i>). The Morgan et al. (1998) equation resulted in the lowest runoff erosion, while the Rauws and Govers (1988) function presented better adjustment to the monitored total sediment production. The Morgan (2001) equation presented the best <i>NSE</i> (Nash–Sutcliffe) values for the simulated sediment production series. These results can help in calculating the coefficient of soil detachment by runoff under low infiltration and high compaction conditions.</p>