This dissertation studies the effect of two practical attributes of structural materials on ultrasonic inspection. The effects of the material microstructure of polycrystalline aluminum alloys on propagating ultrasonic waves and the effects of surface curvature and roughness on ultrasonic scattering and flaw detectability are both addressed.;The relationships between ultrasonic properties (velocity, attenuation, and backscattering) and the microstructure are studied both experimentally and theoretically for rolled aluminum samples with highly elongated grains. Attenuation measurements show a very small anisotropy, whereas backscattering measurements show a very high anisotropy. Existing theories and related extensions are able to explain these physical phenomena. Three approaches are proposed to simultaneously determine grain size and shape. Each approach gives reasonably accurate estimates of grain size and shape.;Three existing models are used to study how cylindrical surfaces affect transducer radiation fields and signal-to-noise ratios through a flaw detection example: the Gauss-Hermite beam model, the Born flaw signal model, and the microstructural response model. The beam model can explain the defocusing effect of cylindrical surfaces. The flaw signal model and the microstructural response model combine to yield estimates of signal-to-noise ratios and the minimum detectable inclusion sizes.;Two time-domain theories that can predict ultrasonic backscattered noise due to a periodically rough entry surface in a pulse/echo immersion test are presented. The first is an exact result and the second is a computationally efficient engineering approximation. Experimental verifications of the theories are presented at both normal incidence and oblique incidence. The predictions are compared to experiment, with good agreement between the theory and the experiment being observed in most cases. Practical implications of the theory for ultrasonic flaw detection and materials characterization are also discussed.