The paper describes a study demonstrating that the screening of a few members of the population and asking them about the distribution of psychiatric symptoms in total population is a very inadequate way of discovering the real prevalence rates. The analysis shows that people report symptoms more amongst those who are socially and geographically close to them and amongst the members of their own sex. The characteristics of the 'reporters' are analysed and the results show that the young, the rich, the highly educated and those belonging to more advanced sections of the society are more prone to reporting symptoms in others. The most interesting finding is that those who have psychiatric symptoms themselves report symptoms in others more than those who are symptom-free. © 1975 Springer-Verlag.