1. Acuity cards are now commonly used to assess the grating acuity of infants, but the procedures vary across clinical settings. To evaluate differences between various acuity card procedures and a more objective preferential looking test of acuity, we compared each of three acuity card procedures with a preferential looking (PL) staircase procedure in toddlers with ocular disorders. 2. In Expt 1, we used PL and two or three variations of the acuity card procedure to measure grating acuity in one eye of 30 toddlers aged 15-30 months expected to have reduced acuity because of structural ocular abnormalities. For the acuity card procedures, observers knew either the stripes' location (random procedure), their relative size (Dobson procedure), or their location and absolute size (informed procedure). For comparison, we tested 38 additional children, aged 16-32 months, with two tests of PL. 3. All three acuity card procedures yielded results which agreed moderately well with PL, with Dobson's procedure agreeing best overall. However, the acuity card procedures agreed with PL within 1/2 octave only 44-67% of the time, whereas two tests of PL agreed within I octave 79% of the time. A comparison of the results from the different acuity card procedures did not correspond to those that would be predicted by observer bias. 4. In Expt 2, we tested interocular acuity differences using PL and the Dobson acuity card procedure in 30 children aged 4-35 months who were suspected to be amblyopic. 5. PL and acuity cards agreed for only half the patients on whether or not there was an interocular difference of at least I octave. A comparison of the results from each procedure to the expected interocular acuity differences based on medical history indicated that the acuity card procedure was either (a) influenced by an observer's expectation of interocular differences or (b) more sensitive for detecting expected interocular differences than is PL. 6. We conclude that acuity cards, used with these procedures, cannot substitute for PL, but can yield moderately similar results in much less time.