Purpose: Although vanishing optotype preferential-looking tasks are commonly used to measure visual acuity (VA), the relative sensitivity of these tests to refractive error is not well understood. To address this issue, we determined the effect of spherical and astigmatic simulated refractive errors on adult VA measures obtained using vanishing optotypes, picture optotypes and Sloan letters. Methods: VA was determined uniocularly for adults under conditions of spherical (0.0–3.0 DS; n = 23) and astigmatic (0.0–3.0 DC at 90° and 180°; n = 20) defocus using the Cardiff Acuity Test (vanishing optotypes), crowded linear Lea Symbols (picture-optotype recognition task) and the Early Treatment of Diabetic Retinopathy Study (ETDRS) letter chart. Results: The Cardiff Acuity Test over-estimated VA compared with the Lea Symbols and ETDRS charts in both focused and defocused conditions. The mean difference between the Cardiff Acuity Test and the ETDRS chart was 0.31 logMAR (95% limits of agreement (LOA) 0.10–0.52 logMAR) in focused conditions and 0.64 logMAR (95% LOA 0.25–1.05 logMAR) with 3D of spherical defocus. Defocus degraded VA on all charts, however there was a significant chart-by-defocus interaction whereby the Cardiff Acuity Test was more resistant to the effects of both spherical (P < 0.0001) and cylindrical (P < 0.001) optical defocus than the recognition acuity tasks at all defocus levels. Conclusion: Although the Cardiff Acuity Test provides an easy method for VA measurement in infants and toddlers, there is a considerable overestimation of VA compared with recognition acuity tasks particularly in the presence of defocus. A simple correction factor (of for example three lines overestimate) cannot be applied to Cardiff acuity measures as there is increasing over-estimation of VA with increasing defocus. Infants with significant refractive error may fall within normal visual acuity ranges for the Cardiff Acuity Test.