The viscoelastic response of polymer-based composite laminates, which may take years to develop in service, must be anticipated and accommodated at the design stage. Accelerated testing is therefore required to allow long-term compliance predictions for composite laminates of arbitrary layup, based solely upon short-term tests. In this study, an accelerated viscoelastic characterization scheme is applied to T300/5208 graphiteepoxy laminates. The viscoelastic response of unidirectional specimens is modeled using the theory developed by Schapery. The transient component of the viscoelastic creep compliance is assumed to follow a power law approximation. A recursive relationship is developed, based upon the Schapery single-integral equation, which allows approximation of a continuous time-varying uniaxial load using discrete steps in stress. The viscoelastic response of T300/5208 graphite-epoxy at 149C to transverse normal and shear stresses is determined using 90-deg and 10-deg off-axis tensile specimens, respectively. parameters In each case the seven viscoelastic material required in the analysis are determined experimentally, using a short-term creep/creep recovery testing cycle. A sensitivity analysis is used to select the appropriate short-term test cycle. It is shown that an accurate measure of the power law exponent is crucial for accurate long-term predictions, and that the calculated value of the power law exponent is very sensitive to slight experimental error in recovery data. Based upon this analysis, a 480/120 minute creep/creep recovery test cycle is selected, and the power law exponent is calculated using creep data. A short-term test cycle selection procedure is proposed, which should provide useful guidelines when other viscoelastic materials are being evaluated. Results from the short-term tests on unidirectional specimens are combined using classical lamination theory to provide long-term predictions for symmetric composite laminates. Experimental measurement of the long-term creep compliance at 149C of two distinct T300/5208 laminates is obtained. A reasonable comparison between theory and experiment is observed at time up to 10 5 minutes. Discrepancies which do exist are believed to be due to an insufficient modeling of biaxial stress interactions, to the accumulation of damage in the form of matrix cracks or voids, and/or to interlaminar shear deformations which may occur due to viscoelastic effects or damage accumulation. / Doctor of Philosophy