Boys classified as nonhyperactive learning-disabled (LD), hyperactive but not LD, and as normal in behavior and achievement were contrasted on a visual search task hypothesized to elicit differences in selective and sustained attention. As task complexity increased, both clinical groups took more trials to reach criterion than the normals, but contrary to expectation, their error patterns were more similar than different. Using a highly similar task, Pribram found that monkeys with frontal lobe lesions did not condition to the reward contingency as well as temporarily lesioned monkeys, who had greater difficulty discovering the correct cue when more than six objects comprised the field. Both LD and hyperactive children who were inefficient at searching out the to-be rewarded cue tended not to condition as well to the reward contingency. Hyperactives gained insight into the features of the game somewhat quicker than LD subjects but did not use this information consistently. The hyperactive and LD groups complained equally of becoming tired but more of the hyperactives wanted to quit and gave evidence in latency data and extraneous responses that they ceased to play the game in a straightforward way. The hyperactives but not the other groups had a negative correlation between accuracy and search latency during the latter phase of the experiment. The hyperactives' lack of “tolerance for a problem” is theorized to account for their poor performance, whereas the LD children's poor performance reflects defects in selective as well as sustained attention.