Parameters such as dimensions, edge parallelism, and edge defect are important in separating good rectangular bars from bad ones. This work proposes a Fourier-optics-based noninvasive quality assessment tool for a rectangular bar ranging from a few micrometers to submillimeters thick. Our design concept relies on a transmissive optical architecture to reduce the effect of an object's angle misalignment that can introduce a significant measurement error. From the far-field diffraction pattern of the bar, the thickness of the bar can be determined from the distance between the two adjacent diffracted order beams. In addition, because all diffracted order beams are aligned, we use the resulting slope to determine edge parallelism. The amount of edge defect on the sample can be investigated by evaluating the distribution of the optical intensity inside diffracted order beams. Our experiment using a 635-nm wavelength laser diode, a 150-mm focal length lens, and an image sensor with 8.4-µm pixel pitch for 22 sample bars with an average thickness of 238 µm shows that our approach can simultaneously evaluate the thickness and the edge parallelism of the bar samples, as well as distinguish nicely edged bars from poorly edged bars. Other key features include low cost, ease of implementation, robustness, and low component counts.