Longshore sediment transport (LST) acts on beach morphodynamics on distinct temporal scales, being fundamental over sediment budget, and, so over the dynamic balance of coastlines. This study determined the most adequate methodology to estimate rates of non-cohesive sediments (from fine sand to gravel) transported by longshore currents, in the surf zone, under different meteoceanographic conditions, at the southern coast of Rio Grande do Sul (RS), Brazil. The methods proposed by CERC (Shore Protection Manual, Coastal Engineering Research Center (CERC), US Army Corps of Engineers Research and Development Center, Coastal and Hydraulics Laboratory, Vicksburg, 1984), Kamphuis (J Waterw Port Coast Ocean Eng 117(6):624-640, 1991), Bayramet al.(Coast Eng 54:700-710, 2007) and Van Rijn (Coast Eng 90:23-33, 2014) were tested through the comparison with data collected with streamer traps, as well as through statistical tests. Van Rijn (2014) was elected the most suitable approach for all beaches studied, followed by CERC (1984). The other two did not perform well for the study area. All models resulted in underestimated rates during high-energy scenarios, and this is attributed to wind negligence or insufficient weight factor in model's calculations, besides the lack of sampled data during very energetic conditions. Wave characteristics, median grain size and beach slope were proven to strongly influence LST, in accordance with the choice of best model overall. LST behavioural response and model's effectiveness will be different for each specific region, especially in what concerns the beach morphodynamic stage. Hence, the method prediction accuracy depends greatly on the suitability of the dataset with respect to the analytical model being used.