Safety testing is a major responsibility of toxicologists. Toxicology is not only a science but also an art. The science of toxicology characterises the toxic potential of a given chemical entity, i.e. the intrinsic property which allows it to react with, and/or to be transformed by, a particular biological system. Based on such scientific data, the art of toxicology has to predict the risk, i.e. the probability that a particular adverse event will occur during a stated period of time or result from a particular challenge. Until now, the science of toxicology has relied almost exclusively on animal tests, the protocols of which are described in directives and regulations. As stated in an Editorial in ATLA (1) the question that toxicologists now have to tackle is, “can non-animal toxicity studies become genuine replacement alternatives …” for assessing risk adequately? Indeed, the science of toxicology has developed, and continues to develop, new approaches (alternatives) to characterise, in well-defined in vitro models (including, for the first time, human models), the toxic potential of chemicals, namely, cytotoxicity, organ-specific effects, modulation of metabolic functions, interference with cell-mediated processes, metabolic activation, etc. But the question remains, what about the art of toxicology? Is it realistic to predict that such new scientific data will, in time, be accepted by regulators for risk evaluation? If these data are to be accepted, we believe that, instead of the present trend towards a regulation-required “protocol toxicology”, toxicologists will have to impose a stepwise decision-tier approach based on the systematic and sequential progression of scientifically justified and rigorously performed investigations, the results of which will be thoroughly and realistically evaluated by experts. It has to be recognised that scientific knowledge has advanced far enough to permit a focus on mechanisms, so that alternatives are fully accepted, no longer as a supplement to a check-list approach, but as a full part of the scientific expertise.