This paper discusses the process through which a powerful multidimensional measure of affect and cognition in relation to adult learning of computing skills was derived from its early theoretical stages to its validation using structural equation modeling. The discussion emphasizes the importance of ensuring a strong substantive base from which to develop reliable items for a measure, as well as the usefulness of gathering qualitative data both in the factor and item design stages to supplement the confirmatory factor analyses. The Computer Anxiety and Learning Measure (CALM) was developed and then validated using a sample of 794 undergraduates in the Sydney (Australia) area. About 70% of the sample came from an English-speaking background, and they represented four university faculties. The original 100 items were reduced to 65 through a combination of exploratory and confirmatory factor analyses and the evaluation of item face validity by three independent raters. Eleven first-order and two second-order factors were subsumed into four subscales. Results from the multimethod techniques used to validate this instrument provide strong support for the validity and reliability of the CALM measure, which appears to have wide applicability for the assessment of anxiety and negative cognitions of adult learners about computer use. Two appendixes contain information on goodness of fit statistics and standardized estimates for the individual CALM items. (Contains 1 figure, 7 tables, and 48 references.) (SLD)