The present paper critically reviews the notion of safety pharmacology. The term "safety pharmacology" covers those studies carried out to evaluate the risk potential of a novel substance on different biological systems. Tests for other activities could more appropriately come under the heading of "general pharmacology". Studies to be included (central nervous system, somatic nervous system, autonomic nervous system, respiratory and cardiovascular systems, and digestive system) have been most comprehensively treated in the Japanese guidelines, but important aspects (effects on higher cognitive function, drug abuse and dependence) are not covered there. On the other hand, the Japanese guidelines provide specific recommendations concerning species, numbers per group, administration route, doses, and control groups. It is important that the tests employed detect bidirectional drug effects and that the tests performed be calibrated in both directions with appropriate reference substances. This requirement is less appropriate for multiparameter procedures. Blind testing could be an advantage. Ethical considerations are important, but the ultimate ethical criterion is the assessment of risk for humans. It is concluded that safety pharmacology studies should not be over-inclusive, but should be performed to the most exacting standards, including Good Laboratory Practice (GLP) compliance. Safety pharmacology data should be available during the planning stage for Phase I studies.