This article concentrates on development of statistical models for prediction of biomass components (above and below ground) of standing trees of Populus deltoides. Twenty seven trees (three each from age one to nine years) were destructively harvested, separated, sorted, sub-sampled, dried to constant weight at 60 C and weighted for biomass components (leaf, twig, branch, bole, stump root, lateral root, fine root). Harvesting in a similar manner, was continued annually up to nine years of tree age and thus in all 27 sampled trees were available for analysis and fitting of models. Diameter at breast height (dbh) alone was a very good predictor of dry weight and accordingly the height was not included in the model. Various functions viz (linear, allometric, logistic, gompertz and chapman-richards), were attempted for dry weight estimation. The linear model, though easiest to fit, suffered from the 'negative estimation problem', specifically for the lower range of explanatory variate. Of the remaining non-linear models, the allometric model out-performed the others on the basis of validation criterions. The value of R-2 ranged from 0.95 to 0.99, for the allometric models fitted on various biomass components. The proposed models can be used for prediction of component wise dry biomass of P. deltoides for a wide range of dbh values (1-50 cm) at one end and can also help farmers in the choice of economical harvest rather than the traditional physical rotation. In addition, they can be used in carbon sequestration studies, which needs complete biomass estimation. (C) 2010 Elsevier Ltd. All rights reserved.