Most antiarrhythmic agents were discovered accidentally. In the last decade, the understanding of the mechanisms of action of agents with electrophysiologic activity has progressed greatly. As a result, it was possible to compute, before the CAST trial, that the agents selected for the trial would not be effective against tachycardias and that the drugs would be unsafe. Extension of these computations to existing Class I agents indicated that they were all poor suppressors of ventricular tachycardia. Furthermore, a Class I agent with an optimal electrophysiologic profile still computes to be a two-edged sword, possessing both antiarrhythmic and proarrhythmic properties. Fortunately, it is possible to conceive of drug profiles that would be purer antiarrhythmic agents. For example, a drug that only upon the development of a tachycardia lengthens action potential duration in a use-dependent manner until the refractory period exceeds the tachycardia cycle length will render continuation of the tachycardia impossible. Recognition of chemicals that have Class IIIa properties with the appropriate kinetics is a challenging task. However, today's microprocessors have become powerful enough to characterize the Class III kinetics. A system that fully automatically screens for effective antiarrhythmic agents is described. It is expected that chemicals selected for optimal basic electrophysiologic properties will yield safer and more effective antiarrhythmic agents.