<p>The concept of scale (size) effect was experimentally studied to establish its impact on the float glass strength influenced by the specimen thickness in four-point bending tests of flat plates of different geometries. The statistical approach to the ultimate strength distribution was used. Fractographic analysis of specimens in bending revealed its dominant fracture mode as the propagation of a main macrocrack from the vulnerable edge defect. Experimental data were represented by the box plots for the four groups of examined specimens, including statistical factors such as mean value, median, and lower and upper quartiles, as well as results that may be erroneous and considered outliers. Such anomalous results and their elimination were verified with the Chauvenet and Smirnov statistical criteria, considerably reducing the empirical data scattering. The statistical distributions of glass strength displayed the uniformity level of existing defects, which showed up in data scatter patterns. The widest and narrowest scatter was found for the specimens 3.85 and 6.0 mm thick, respectively, indicating the outgoing quality of flat glass, the effect of process, production, and machining factors. The bimodal strength distribution was established for the specimens 3.85 mm thick, induced by two types of determining defects, located at the edge and in the near-surface cracked glass layer. The diagram displaying a direct scale effect in determining the ultimate strength is presented for the specimens 3.85–8.2 mm thick; this effect should be considered in engineering design for reliable evaluation of performance and service life of critical structural float glass elements.</p>