A key factor limiting the use of fracture mechanics in describing the compressive behaviour of brittle engineering materials is the analytical complexity associated with the need to consider two-dimensional flaw geometries. Researchers have observed that linear elastic (LE) stress fields surrounding voids in compression appear to have predictive value regarding compressive fracture. A theoretical framework for the interpretation of LE stress fields towards compressive fracture, however, remains outstanding. Herein fractal void analysis is proposed to enable interpretation of LE stress fields in the context of Mode I compressive fracture, yielding significant theoretical guidance for this intractable problem. Fractal void analysis consists of considering void surfaces as fractal, and applying what is termed the “small flaw assumption” to analyze stress and energy release rates at the tips of self-affine “surface irregularities”. Fractal void analysis is first applied to two-dimensional LE systems. It is verified against finite element and linear elasticity solutions, and found to account for shape and size effects (key prior limitations). Significantly, two-dimensional fractal void analysis yields efficient KI estimates; peak KI values for radial line cracks emanating from circular voids are calculated to 74% of benchmark numerical values via spreadsheet. Fractal void analysis is then applied to three-dimensional LE systems, where it predicts images of “needle cracks” recently contributed to the literature and suggests two possible mechanisms for planar crack formation. Conclusions from fractal void analysis then support a theoretical analysis of compressive fracture in cementitious materials. The importance of macropore and aggregate distributions in crack initiation is shown. The behaviour of cylinder and masonry prism compression specimens is also considered: successful prediction of cone-shaped fracture patterns in cylinder specimens with length to diameter ratios L/d ≤ 2 is presented, and a discussion of the “lateral tensile splitting theory” in understanding masonry prism fracture is provided. Several areas of potential future research are identified. Theoretical areas include analyses including accounting for mixed-mode fracture and further study of planar crack formation from spheroidal void surfaces. Outstanding experimental work includes correlating test results from materials with simple void distributions to fractal void analysis results.