The concept of cognitive holding power is synthesized from theories of settings and of cognitive structures and is conceptualized as a characteristic of a learning setting that presses students into different kinds of cognitive activity. Settings which press students into using first‐ or second‐order cognitive procedures are regarded as having first‐ or second‐order cognitive holding power. The development of an instrument to measure these two dimensions of cognitive holding power is outlined. The independence of the dimensions, their reliabilities and validity, and factor structures are examined. Each dimension was found to have high reliability across vocational education and high school settings, and each was correlated as predicted with other classroom variables. The potential contribution of this research to understanding the relationship between different approaches to the teaching of problem solving and the ability to undertake problem‐solving transfer tasks is outlined.