This study sought to analyze the dimensionality of a set of items that measure work-related physical demand, their internal consistency and their test-retest reliability. We carried out an exploratory factorial analysis of ten physical demand items in two populations, totaling 1,070 workers from the footwear industry and from urban cleaning, using weighted least squares means and adjusted variance with Promax oblique rotation. We measure internal consistency using Cronbach's alpha. The test-retest used a subsample of 121 workers, using quadratic weighted kappa. We identified two dimensions in urban cleaning and three in the footwear industry: (1) dynamic work with anomalous postures; (2) work that requires handling material; and (3) work that demands repetitive motions. Reliability was excellent for upright work posture, good for walking, moderate for lifting loads, squatting, rotated trunk and only regular for repetitive motions and arms above shoulders. The dimensions we identified through factorial analysis are compatible with the physical demands of urban cleaning and the footwear industry and, in general, showed good internal consistency. The 10 items' reliability was compatible with the literature, showing excellent to good reliability for general posture, moderate for specific body segment posture and regular for repetitiveness. Results reveal the specificity of each occupational group, which suggests a dimensionality of physical demand items, supporting their use for measuring physical exposure.