Bernstein (1969, 1979) has argued that Skolov's (1963) model is not sufficient to account for the data concerning orienting response (OR) elicitation to stimulus change. He has argued that following the registration of stimulus mismatch, an additional stage of stimulus evaluation occurs and that only stimuli which represent change and are judged as 'significant' elicit ORs. The present paper takes issue with this assertion. It is argued that the data on which the 'significance hypothesis' was originally based are not replicable and that the argument, as it was formulated originally, is tautological. It is also argued that individual differences in OR responsiveness are more parsimoniously explained in terms of individual response stereotypy and that differential responsiveness to different sorts of stimulus change poses problems for the significance hypothesis. Evidence is also reviewed which suggests that the relationship between stimulus change and stimulus significance is additive rather than multiplicative as suggested by Bernstein, i.e., stimulus change and stimulus significance are each sufficient for OR elicitation. Finally, it is concluded that the concept of 'significance', introduced operationally, may be of value in terms of examining the role of the OR in information processing.