The role of element proximity in perceptual grouping was examined in tasks requmng speeded discrimination of two-element visual patterns. Grouping of two elements was defined as the failure of attention to be focused on one element selectively in filtering tasks where only that one element was relevant to the discrimination. Failure of selective attention was measured by the degree of interference caused by variation of the irrelevant element. Grouping was shown to diminish monotonically as the spacing between two elements was increased. At a given spacing, grouping could be reduced or eliminated by the introduction of a third element into the stimulus field, presumably because the addition of this element triggered a reorganization of the perceptual field into a new grouping structure. Grouping appeared to facilitate performance on condensation tasks requiring distributed attention, to the degree that the condensation tasks were actually easier than the filtering tasks at close proximities. Paradoxically, for some tasks, moving an irrelevant element away from a relevant one actually impaired performance, suggesting that paying attention to irrelevant information could be beneficial. This result, if generalizable, suggests that grouping be conceptualized not as an automatic process under preattentive control but as an optional process under strategic control. This paper is concerned with two aspects of the problem 01 perceptual organization of visual arrays (Wertheimer, 1923). The first is the degree to which perceptual organization is under the control of specitic parameters of the stimulus array, as opposed to cognitive (strategic) factors within the observer. The second is the establishment of experimental measures of organizational processes. This latter issue is much more than simply a methodological one, for in order to measure organization a theory is required that explains just what organization means, and how and where organization enters into the processing of perceptual information.The Gestalt "laws" of course represented an attempt to answer the first question. Certain stimulus variables. some well defined (proximity) and some not so well defined (similarity), can be shown to have clear effects on the wayan observer reports his perception of stimulus arrays. The conclusions one can draw from these demonstrations. as compelling as the perceptual effects may be, are limited, however. to the domain of phenomenology and say little about the underlying processes that shape our perceptual experiences. What is required are performance measures. rather than subjective reports. of perceptual organization. Pomerantz and Garner (1973) argued that if organization is more than an epiphenomenon. there