To compare the effects of "simulated" and "real" spherical and astigmatic defocus on visual acuity (VA). VA was determined with letter charts that were blurred by calculated spherical or astigmatic defocus (simulated defocus) or were seen through spherical or astigmatic trial lenses (real defocus). Defocus was simulated using ZEMAX and the Liou-Brennan eye model. Nine subjects participated [mean age, 27.2 ± 1.8 years; logarithm of the minimum angle of resolution (logMAR), -0.1]. Three different experiments were conducted in which VA was reduced by 20% (logMAR 0.0), 50% (logMAR 0.2), or 75% (logMAR 0.5) by either (1) imposing positive spherical defocus, (2) imposing positive and negative astigmatic defocus in three axes (0, 45, and 90°), and (3) imposing cross-cylinder defocus in the same three axes as in (2). Experiment (1): there were only minor differences in VA with simulated and real positive spherical defocus. Experiment (2): simulated astigmatic defocus reduced VA twice as much as real astigmatic defocus in all tested axes (p < 0.01 in all cases). Experiment (3): simulated cross-cylinder defocus reduced VA much more than real cross-cylinder defocus (p < 0.01 in all cases), similarly for all three tested axes. The visual system appears more tolerant against "real" spherical, astigmatic, and cross-cylinder defocus than against "simulated" blur. Possible reasons could be (1) limitations in the modeling procedures to simulate defocus, (2) higher ocular aberrations, and (3) fluctuations of accommodation. However, the two optical explanations (2) and (3) cannot account for the magnitude of the effect, and (1) was carefully analyzed. It is proposed that something may be special about the visual processing of real astigmatic and cross-cylinder defocus-because they have less effect on VA than simulations predict.