There is a deep-rooted belief among some professional workers in the field of educational theory and practice that the optimal distribution of errors or marks should correspond to a distribution pattern according to Gauss’ curve, i. e. a normal distribution pattern. The author of this article points out that this distribution pattern is only one of the possible patterns and its promotion to the status of a general distribution pattern has a number of negative consequences. It is particularly unfair to some pupils, it leads to the confusion of pedagogical and psychological viewpoints in appraising pupils, it decreases the validity of examinations. The appraisal of pupils ought to proceed from educational aims and from the control concerning the fulfilment of normatively set tasks, whose feasibility of fulfilment must be verified by practical tests and examinations. There is no reason, however, for arranging these examinations in such a way as to make their results correspond to the normal distribution pattern.