Experiments in "hypnosis" are criticized as having failed to control significant variables. These variables are as follows: (1) In most experiments, a single group of "good" hypnotic subjects were tested under both "trance" and non-trance conditions. In some of these experiments it appears that the "good" hypnotic subjects, in "striving" to comply with what they correctly or incorrectly surmised were the expectations or wishes of the experimenter, purposively gave their best possible performance under the "hypnosis" condition and purposively gave an inferior performance under the control condition. This factor was not controlled in other studies which used this experimental design. (2) In studies employing independent experimental and control groups, the "trance"-waking comparisons were biased by the following factors: The experimental group, but not the control group, had participated in preliminary "training " sessions in which it had received practice in performing tasks which were similar to or identical with the criterion tasks. The experimental group consisted of highly selected individuals who were judged to be "good" hypnotic subjects. In many instances the control group was selected from individuals rated as "poor" hypnotic subjects. A series of studies suggests that, prior to participation in "hypnosis" experiments, potentially "good" hypnotic subjects tend to comply with the requests of a respected person and tend to be willing to carry out activities involving imagery and fantasy, and potentially "poor" hypnotic subjects tend to resist domination from others and tend to be unwilling to perform imaginational activities. Personality differences between subjects may thus have been confounded with the treatment effects. A friendly relationship between subjects and experimenter was at times formed in the preliminary "training" sessions which were held with the experimental group but not with the control group. This relationship was then strengthened during the treatment sessions when the "hypnosis" group received more of the experimenter''s time and interest than the control group. Since the experimental group had interacted more extensively with the experimenter, and had possibly formed a closer relationship with him, it may have been more willing (with or without "hypnosis") to comply with his wishes or suggestions. To the experimental group the situation was explicitly defined as "hypnosis" and explicitly or implicitly defined as an important and unusual situation in which interesting events were to be expected. In addition, special efforts were usually made to motivate the experimental group to perform well on the criterion tasks. In contrast, the situation was generally defined to the control subjects as an ordinary laboratory situation and no effort was made to arouse their interest or to motivate them to perform maximally on the criteria. Discrepancies in interest, expectancies, and motivation may have been sufficient to produce some of the observed differences in behavior, independently of the presence or absence of "hypnotic trance." An experimental design which controls these variables is presented and recommended for use in "hypnosis" research.