Background This study was motivated by the ubiquity and apparent usefulness of general epistemological development schemes, notably that of William J. Perry, Jr., in engineering education, but also by limitations that derive from their generality. Purpose/Hypothesis Empirical data were used to articulate engineering students' epistemological views on the role of mathematical methods in engineering and to explore the fit of a stage-based developmental model to those data. Design/Method Data included interviews, think-aloud protocols, and classroom observations over a one-year period. Ten undergraduates and four instructors in a civil engineering program participated. A grounded-theory approach was used to identify levels of epistemological views. Perry's scheme provided a starting framework. Skeptical reverence, the view veteran engineers hold regarding mathematics in engineering, which was previously identified by the author, was taken as a normative endpoint. All data were coded by view level and various contexts to detect students' epistemological developmental patterns. Results This article proposes three categories of engineering students' views on the role of mathematical methods in engineering: dualism, integrating, and relativism. Dualism and relativism reflect elements of Perry's general categories, but integrating, a new category, diverges significantly from Perry's middle category of multiplicity. No evidence supported a stage-based developmental model. Conclusions This empirically based scheme, while exploratory, provides further evidence that epistemological development differs across disciplines, and offers four levels of epistemological views held by engineering students on the role of mathematics in engineering. Conjectures about how to promote engineering students' epistemological development, based on classroom observations, are also offered.