This double masked study compares the cycloplegic effects of tropicamide 1% and cyclopentolate 1% in 20 nonstrabismic, nonamblyopic, hyperopic 6- to 12-year-old children with a mean refractive error = +1.48 +/- 1.10 diopters (D). Unlike previous studies which used only amplitude of accommodation to measure the depth of cycloplegia, this study compares refractive error as determined by retinoscopy, distance subjective refraction, and distance autorefraction (Canon R-1). In addition, we compare the amplitude of accommodation as measured by subjective push-up and objective autorefraction methods. There is no statistically significant difference between cyclopentolate and tropicamide for either cycloplegic retinoscopy or distance subjective refraction. Autorefraction measurement of refractive error shows a statistically significant but clinically unimportant bias (0.14 +/- 0.30 D) toward more hyperopia with cyclopentolate. Both drops reveal latent hyperopia, and the mean latencies are not statistically different between the two cycloplegic agents. Latent hyperopia is not systematically related to the degree of hyperopia after tropicamide, but this relation is significant after cyclopentolate. No differences were found between refractive results with either agent at 30 min compared to 60 min after drop instillation. When measured objectively with the autorefractor, accommodation is inhibited more effectively by cyclopentolate than by tropicamide. Our results suggest that although tropicamide is not as effective as cyclopentolate in inhibiting accommodation it is, nevertheless, a useful cycloplegic agent for measuring distance refractive error of low to moderate hyperopia in school-aged children.