This study tested Albert Ellis' Rational Emotive Therapy (RET) theory which predicts that cognitive beliefs, not the stimulus situation, generate human emotions. According to RET, emotions created by rational beliefs are adaptive, while irrational beliefs result in an unadaptive anxiety level. Results demonstrated that at high levels of problem relevance there was (1) a significantly greater GSR in direct response to the stimulus situation, and also to irrational statements, than to rational and control statements, and (2) no significant difference between rational and neutral control statements. The authors argue that these results are more parsimoniously explained by conditioning theory than by RET theory.