New experimental procedures can sometimes reduce the numbers of animals used, particularly in the testing of new compounds for toxicity. For instance, new chemical compounds are now routinely screened in cell cultures, and if they are found to be toxic they are not given to animals. But cell cultures cannot replace animals. If a compound proves innocuous in a cell culture, it must still be tested in an animal to gauge its effects in a complex organism. Similarly, computer models and other nonanimal alternatives can at most supplement, rather than replace, animal experimentation.