AbstractThe environmental stress crack resistance of polyethylene is measured most conveniently by the bent‐strip method described in ASTM D‐1693‐70. The method has, however, suffered from poor reproducibility of test results. In this work we have carefully studied the effect of individual experimental variables that enter the test method. The effects considered are: the thermal history of test specimens; uncontrolled variations in the test procedures such as the sample holder width, the specimen thickness, and the stress relaxation occurring during the test; and the variability in the test results arising from the statistical nature of fracture process. It is concluded that each of the following four factors, when not properly controlled, could contribute an apparently random variation of a factor of two or more in the test results. These are (1) the aging of polyethylene specimens at room temperature, (2) the width of the specimen holder, (3) the dimensions of the tools employed for bending and transferring specimens, and (4) statistical fluctuations of the final results arising from the limited number of specimens tested and also from the stress relaxation occurring in the specimen during the test. By carefully controlling these and other variables, it should be possible to design a test procedure which is expected to yield results of a known precision and can serve even as a research tool.