Woodland managers in northern Chile require efficient means of estimating fruit production to improve their ability to sustainably manage mesquite tree ecosystem. We assessed the accuracy of the indirect visual count (VC) method for quantifying fruit production of two endangered mesquite trees, tamarugo (Prosopis tamarugo) and algarrobo (Prosopis alba), by comparing it with the fruit trap (FT) method, during three fruit dissemination periods. We studied two natural populations of trees of P. tamarugo and P. alba in the Pampa del Tamarugal ecosystem, Atacama Desert, northern Chile. We placed FTs beneath the crowns of the trees to measure the fruit crop during the mast dissemination period. The VC method was performed at the same trees used for FT. We used linear regressions and coefficient of variation to assess the relationship between the two methods. There was a positive correlation between both methodologies in all three periods for P. alba, but the relationship for P. tamarugo was not statistically significant. Analysis of the variability in fruit production estimates using the FT method showed significant interannual differences for P. tamarugo and significant differences among individuals for both species. The VC method had better results for fruit production in P. alba, whereas the use of the FT method is more suitable for P. tamarugo.