[en] Fracture tests are conducted usually out of a conviction (sometimes only vaguely defined) that they will guarantee a certain level of protecton from metal failure. Qualitative tests, such as the Charpy V-notch, produce results which cannot be rigorously related to a measure of fracture tolerance: rather, they indicate metal quality so that fracture tolerance may be inferred. Quantitative tests on the other hand provide parameters which may be used directly in equations to determine the likelihood of fracture. Both types of tests have limitations which should be understood: the paper tries to provide guidance on the relative merits of either approach for a particular purpose, and gives an insight into near-future test methods which will extend the range of usefullness of quantitative tests. (author)