Randomization of animals by computer program for toxicity studies.

Evaluations of toxic potential of drugs and chemicals often involve statistical comparisons of effects between treatment groups. Such comparisons are valid and permit elucidation of spontaneous versus treatment effects only if the sampling population is obtained without selection bias. Selection bias is routinely minimized or controlled by random sampling or unbiased allocation of animals to treatment groups. A systematic approach using a computerized procedure is described that fulfills these requirements in a simple and efficient fashion.