Digital elevation models (DEMs) are essential in many professional areas as they produce georeferenced elevation data that are critical for a wide range of studies, computations, decision-making processes, and derived products. Quality control thus becomes necessary to quantify the accuracy of the information provided. We assessed the accuracy of elevation data estimated by DEMs derived from LiDAR data representing diverse land cover types. For this purpose, we used the FUSION software and explored variations in accuracy in relation to the following factors: input data, interpolation methods, terrain slope, heterogeneity of land cover, and LiDAR point density. We selected and measured 1157 checkpoints by using total station and GPS techniques and following a stratified random design in order to validate the LiDAR-derived DEMs. We used robust estimators, nonparametric tests, and analysis of variance to examine the elevation errors. The study findings showed the following: (1) using the full set of LiDAR returns did not improve elevation accuracy relative to using the last-return data set; (2) using the minimum switch for interpolation did not improve accuracy compared to the default behavior of the interpolator; (3) land cover and slope significantly affected accuracy; (4) DEMs tended to underestimate elevation; and (5) the mean density of the returns classified as ground was significantly affected by land cover and slope factors.