With an aim to improve vocational interest assessments geared toward the Science, Technology, Engineering, and Mathematics (STEM) areas, we developed a new assessment by incorporating occupational complexity levels. Occupations which correspond to Holland's realistic and investigative themes were identified together with their complexity levels using the information in the Dictionary of Holland Occupational Codes. How the required skills, abilities, and work activities differ in occupations with different complexity levels was identified using the Occupational Information Network. Two studies (N = 691 total) were conducted to assess reliability and validity of a measure of STEM Interest Complexity assessing the level of interests for STEM, which was developed based on the information obtained from these occupation databases. Construct validity was demonstrated with factor analyses and based on associations with traditional interest assessments, cognitive abilities, and theoretically-related constructs. Concurrent criterion-related validity was demonstrated based on associations with attachment to STEM, intentions to persist in and further pursue a STEM field, intentions to choose a complex occupation, college major satisfaction, and STEM-related GPA. Dominance analyses indicated that the relative contribution of the STEM Interest Complexity measure exceeded that of traditional interest assessments in the prediction of criteria. The usefulness of integrating a dimension of level and consulting occupation information databases in developing vocational interest assessments are discussed. (C) 2011 Elsevier Inc. All rights reserved.