Subjects were presented with 36 opinion statements ostensibly made by residents of two towns (one large, one small) confronted with the building of a new nuclear power station in the vicinity. There were an equal number of pro- and anti-nuclear opinions for both towns (nine for each combination) so that there was no relationship between attitudes and town. The probabilistic belief that small towns would contain a higher proportion of antinuclear residents, was predicted to produce an ‘illusory correlation’ confirming the belief. This was strongly supported. Findings were discussed in terms of the literatures on illusory correlation, stereotyping and attitudes.