Although strength data for glass is available in the handbooks and reference data is given in the building norms such as the NEN 2608, the question is if these results are based on statistically significant sets of data. To determine the strength of glass in bending a large tests series of 180 specimens were prepared. These were cut from a single annealed glass plate, and identically processed on a single cutting and grinding line. One third of these were tested in the annealed condition, one third heat strengthened and one third was fully tempered. All specimens were tested in four point bending. Weibull analyses were made of the results. It is concluded that the strengths determined here are considerably lower than usually given in the literature and in the building norms. In the discussion a number of reasons for this are explained and the relevance of this to engineering is given.