Regression models relating variables derived from airborne laser scanning to the amount of biomass of potential logging residues (PLR) were estimated for 147 sample plots (200 m(2)) measured in mature boreal forest. Logging residues were in this study defined as branches and tops of all trees. The base model explained 86% of the variation, and when the data were stratified into two strata according to site quality, the stratum-specific models accounted for 88% and 77% of the variation in PLR on poor sites and on good sites, respectively. Effect of tree species composition were assessed by including the proportion of Norway spruce as potential explanatory variable and this extended model explained 87% of the variability. The estimated models were validated using two datasets, one comprising 120 sample plots (200 m(2)), and the other consisting of 25 measured stands. The ground observations in the stand dataset were based on data collected by a harvester. The validation of the overall model gave an RMSE value of 27.6% and 22.2% of the mean value in the plot and stand data, respectively. The stratum-specific models gave RMSE values of 25.3 and 22.5% in the plot validation, and 24.3 and 17.6% in the stand validation. Including proportion of spruce in the model gave an RMSE of 23.2% in the plot validation and 21.3% in the stand validation. The study shows that PLR in a boreal forest can be estimated by airborne laser scanning with accuracy comparable to those obtained when estimating e.g. stand volume. (C) 2011 Elsevier Ltd. All rights reserved.