Two experiments are reported on the effect of negation on reasoning with conditional rules. In Expt. I significantly more affirmative than negative statements were denied by a valid inference (modus tollens), and significantly more negative than affirmative statements were affirmed by a fallacious inference (affirmation of the consequent). Subjects made their inferences by choosing from a list of possible conclusions, so that, for example, an affirmative statement, p, could be denied by selecting the negative conclusion, not p. It was hypothesized that the greater difficulty experienced by subjects in the denial of not p by a choice of p might be due to the additional intermediate step of double negation. i.e. not not p, which is involved. The results of Expt. II suggested, however, that the difficulty lies in subjects' inability to infer that a negative statement is false, rather than in the process of transforming a double negative into an affirmative.