The attempt to create a science of politics, so popular in the 1950s and 1960s, has met with heavy criticism in later years. One line of attack has been that the claims of the behaviouralists to be value-free in their research were actually false claims, so that the result was to promote the author’s values without any admission by the author that this was being done and without any reasoned normative arguments to justify the values. A second line of attack has been the claim that the values secretly promoted by the behaviouralists and model-makers are undesirable values. A third and much more basic line of attack has been the claim that it is a philosophical error to suppose that political behaviour can be properly explained in terms of empirical generalizations similar in form to those of the natural sciences. Each of these lines of criticism will now be examined.