Geomorphology, like most scientific disciplines, has evolved from a descriptive stage through quantification to modelling of processes and outcomes. Indeed, high-resolution modelling has been termed the “third branch of science” for its ability to generate high-quality hypotheses, testable in turn through field mapping and laboratory experimentation. I teach the principles and basic applications of modelling using microcomputer spreadsheets. These serve as “transparent boxes” in which graphical output is generated in response to numerical input, and more importantly, in which the modeler (student) can see the equations and order of steps (unlike traditional “black box” models). The technology is widely available and familiarity with it can be assumed, required, or taught at the undergraduate level. Modelling can be taught through a series of spreadsheet exercises in which students, while learning topical material in a traditional sequence, progress from a simple, “cookbook” exercise through more realistic, process-based exercises to the level of addressing a previously unstudied research question. At each step the model can be used to generate testable hypotheses. I have worked with scarp evolution, stream longitudinal profiles, and glacier reconstructions, and plan to include eolian, soil development, mass wasting, and coastal processes in additional exercises. The first year of use of this technique in a junior-level Geomorphology course clarified the strengths and weaknesses of the modelling approach. The strengths involve students learning the capabilities and flaws of modelling, a stronger understanding of causeeffect relationships, and the effects of time on landforms. The weaknesses involve time commitment, supervision in model implementation, and limitations to student ability to independently pursue openended research.