We report a series of experiments in which we assess depth discrimination performance in adults and children using a disparity-balanced target configuration to avoid the effects of anticipatory vergence eye movements. In our first study we found that children outperformed adults by a substantial margin, and the adults were consistently near chance. This was surprising given that we initially tested naïve adults to provide a benchmark for the children’s data, and all observers met the criterion for stereoacuity. In subsequent experiments we recruited groups of inexperienced adult observers and assessed the role of a wide range of spatial and temporal factors in this apparent deficit. We found that the adult performance remained poor in spite of changes to the stimulus layout, exposure duration, and spatial scale. The only manipulations that improved performance were those that limited the binocular disparity to a single sign. We conclude that these data reflect a form of involuntary disparity pooling that makes it difficult for naïve observers to judge depth from disparity from multiple targets. The absence of this effect in children likely reflects the late maturation of global processes and depth cue integration.