Hence, the identification of occupational disease cannot be allowed to rest on epidemiology alone — in effect, to a policy of ‘counting the victims’. It must be backed by experimentation and other methods of predicting the likely health impact on workers of new chemicals, processes or technologies.