This research reports a comparison of expressed beliefs about intelligence and considers the question of what behaviours are rated as intelligent in terms of relevance to the lives of three groups: primary school, high school and college students. Each person was asked to rate a set of 26 intelligence test items chosen from four popular scales (Stanford Binet, WAIS‐R, WISC‐R, and Raven's Progressive Matrices) on relevance to measuring intelligent behaviour, and on perceived difficulty after they had performed these tasks. The relevance ratings were taken as reflections of the group's concept of intelligence. The correlations of relevance ratings with perceived difficulty and with performance were negligible. This suggests that group differences in defining what constitutes intelligent behaviour across the three groups are not reflections of their own abilities in these tasks. Three types of skill were identified in each group by factor analytic techniques: Non‐verbal reasoning, verbal reasoning and retrieval of information. Primary school children rated the verbal reasoning skill as most relevant to measuring intelligence, the high school students treated all three as equally relevant, and the college students rated the non‐verbal reasoning skill to be the most relevant. This pattern of results is attributed to different task demands at primary schools, high schools, and colleges and to the changing basis on which people are differe