The structural performance, reliability and durability of polymer composites can now be correlated with three generic classes of internal defects. The first generic class of chemical structure defects (size 10-100Å) that control critical design properties such as glass transition T_g , moisture absorption, and dimensional changes can be controlled by chemical analysis of raw materials prior to manufacture. A second generic class of manufacturing defects (size greater than l0μm) include inclusions, voids and debonds which are related to manufacturing process control and recognized by ultrasonics, optical scanning and other techniques sensitive to interfacial imperfections. The interaction of these two classes of intrinsic defects with environmental and mechanical stresses produces a third class of macroscopic fatigue defects such as interconnected microcracks and macroscopic crack growth which can be detected by visual inspection and ultrasonic emission. The recognition of intrinsic structural defects, and their contributions to polymer composite reliability, represents an important extension in the analytic modeling and reliability predictions for structural polymers, adhesively bonded metals and high strength fiber reinforced composites in which the physical chemistry parameters appear as primary control variables. This discussion introduces and discusses combined deterministic/statistical models for polymer composite reliability. The molecular process which determines the relation between environmental condition and macroscopic structural effect is detailed within such models and provides important criteria for chemical and manufacturing optimization of polymer composite reliability. Experimental data of aging effects on the statistical strength distributions of structure polymers, metal-to-metal joints and reinforced composites are examined and compared with model· predictions.