The creep and creep recovery behavior of a random fiber composite (SMC-R50) at elevated temperature and constant humidity were investigated experimentally and theoretically. The short time creep response for four constant stress levels at each of four selected temperature levels was experimentally determined. It was found that repeatable results can be obtained by applying a mechanical conditioning prior to each creep and creep recovery test. Creep data were modelled using the Findley equation which contains three parameters, ε_O (the instantaneous creep response), m (the amplitude of transient creep), and n (the time exponent). It was found that the time exponent is a function of time but approaches to an asymptotic value when the duration of creep is long. Thus, at a constant temperature level, one long-time creep test and four short-time creep tests were conducted. The long-time creep results were used to determine the proper time exponent n. The short time creep data for constant load were used to determine the Findley parameters ε_O and m. It was found that the Findley equation represented the creep results very accurately. Based on the short-time creep results, the Findley equation was used to predict the long time creep response and the creep response due to multiple step loadings. Five long time creep experiments were conducted. Four of them were 10,000 minutes long and were conducted at the same stress level (6,510 psi) but different temperature levels. The fifth creep experiment was conducted at 5,425 psi and 185°F over a three week period. Three multiple step creep experiments were conducted. These tests were of different load steps and durations. In all cases, it was found that the Findley equation predicted both long time creep response and multiple step creep response very accurately. Since repeatable results were obtained from conditioned specimens, the test results were compared to experimental data obtained from unconditioned specimens. It was found that experimental results of the conditioned specimens fell within the scatter band of the data for the unconditioned specimens. A free energy based failure criterion (proposed by Reiner and Weisenberg) was coupled with the Findley equation to predict the creep rupture time of SMC-R50. It was found that the critical free energy at the time of failure is temperature dependent. For a constant temperature, the critical free energy required for rupture is essentially a constant. It is also concluded that, for limited data, the Reiner-Weisenberg failure criterion provj.dl!S overall good prediction of the time to failure for SMC-R50. / Ph. D.