The response criteria problem in annoyance surveys, particularly with respect to prognosis, is dealt with in this report. The criteria problem is illustrated with data from self-rating questions in a survey conducted in five areas with different exposures to aircraft noise. The results show that the respondents' response criteria cannot be presupposed to be independent of the exposure conditions. If one wants to measure mean degree of annoyance evoked by an environmental agent in different areas, calibrated scales must be used. It was shown that the response criteria variation in itself may be used as an indicator of the exposure condition. The simultaneous measurement of response categories and stimulus reactions is recommended in annoyance surveys.