To assess the impacts of a low-intensity selective timber harvest on a palm community in Belize, we mapped logging infrastructure (i.e., roads, log landings, skid trails, and stumps) and measured palm regeneration 1 year after a timber harvest carried out using reduced-impact logging (RIL) practices. We sampled palms across a gradient of increasing harvest impact severity from areas not directly affected by logging, in felling gaps, on secondary and primary skid trails, and on log landings. We used generalised linear mixed-effect models fitted in a Bayesian framework and applied a non-metric multi-dimensional scaling of the ecological distances between sites to evaluate differences in palm seedling regeneration density and species composition, respectively. The harvest of an average of 2.5 trees ha−1 caused 0.4% of the forest to be converted to log landings, 0.7% and 3.6% to roads and skid trails, and 2.3% to felling gaps, which left 93.0% of the 350 ha harvest block with no direct impacts of logging. The difference in abundance and species composition of palm regeneration in unlogged areas compared to felling gaps and skid trails was small, but log landings had markedly lower densities. These results highlight that the impacts of selective logging are minor at least where harvest intensities are low and RIL practices are employed. If further reductions in canopy opening and soil disturbance are desired, we recommend that logs be cable-yarded (i.e., winched) the final 20 m to skid trails instead of driving to the tree stumps. We estimate that implementation of this practice would reduce total skid trail coverage from 3.6% to 2.9% and overall forest disturbance from 7.0% to 6.3%. However, further reductions in disturbance might be inimical to the maintenance of palms and will certainly not favour regeneration of light-demanding commercial timber species (e.g., Swietenia macrophylla).