Wood with a uniform moisture distribution is known to have different mechanical properties compared to wood with a non-uniform moisture distribution. Moisture gradients are likely to develop in full-size members tested in the In-Grade Testing Program and might therefore affect the test results. The purpose of this study was to mathematically model the effect of desorption moisture gradients on the stiffness and strength of yellow-poplar beams. An additional objective was to experimentally determine gradient effects in yellow-poplar beams. Three-dimensional finite-element modeling was employed and several subsidiary models were developed. Among these was a three-parameter segmented model for fitting digitized tension and compression stress-strain curves. Unlike previous models (such as the Ramberg-Osgood model), this model has a linear slope up to the point approximately corresponding to the proportional limit. A methodology was also devised whereby most hardwood and softwood elastic constants can be estimated at any moisture content. Data are required at one moisture content. Equilibrated uniaxial testing was conducted at four moisture contents to acquire data for the finite-element model. It was found that the longitudinal Young's moduli in tension and compression were approximately equal at 6% and 18% moisture content; the compression modulus was greater at 12%, but the tension modulus was greater for green specimens. Intersection points for tension and compression mechanical properties may be different. Tests of small clear yellow-poplar beams indicated that moisture gradients induced at 12% equilibrium moisture content had little effect on the modulus of rupture up to 19% average moisture content. At higher moisture contents, gradient-containing beams were significantly stronger than equilibrated beams when comparisons were made at identical moisture contents. Modulus of elasticity data exhibited a similar trend, although differences between equilibrated and non-equilibrated beams were observed below 19% moisture content. The finite-element program was moderately successful in predicting the effects of moisture gradients on the strength and stiffness of yellow-poplar beams. Computer time and storage constraints limited the accuracy of the solutions. Predicted trends were verified by the experimental data. Modeling of full-size lumber indicated that significant moisture gradients will likely influence the stiffness and strength of higher quality lumber. / Ph. D.