PURPOSE. To study the development of sensitivity to first- and second-order gratings by measuring thresholds for the detection of motion in 3-month-olds, 5-year-olds, and adults. METHODS. Subjects (n = 24/age) saw two 15 x 15º vertical 0.5 c/deg sine-wave gratings that were separated horizontally by a 5 gap. The gratings were added to (first-order condition) or multiplied with (second-order condition) binary noise. Randomly, on each trial one of the gratings was stationary and the other moved outward at 6 deg/sec. Amplitude modulation was varied over trials. For babies, we used the method-of-constant stimuli and data from 16 trials per baby to calculate two group thresholds, one for first-order gratings and one for second-order gratings. On each trial, a trained tester who was unaware of the side of motion decided whether the moving grating was on the left or right based on any reliable cues provided by a baby (direction of first look, of longest look, etc.). The group thresholds from babies were compared to the mean individual thresholds from 5-year-olds and adults who, on each trial, indicated which side had the moving stripes. RESULTS. Thresholds for infants were far more immature for second-order gratings (2.5 times worse than adults) than for first-order gratings (1.09 times worse than adults). However, by 5 years of age, thresholds were close to adult levels (1.03 and 1.09 times worse than adults for first-order and second-order gratings, respectively) and were no more immature for second-order than for first-order gratings (p > 0.05). CONCLUSIONS. During early infancy, the neural mechanisms that detect second-order gratings are especially immature compared to those that detect first-order gratings. By 5 years of age, at least under the present testing conditions, mechanisms detecting second-order gratings are almost adult-like and are no less mature than those detecting first-order gratings.