This study investigated the Gestalt law of closure in baboons. Using a computer-controlled self-testing procedure, we trained baboons (Papio papio) to discriminate open versus closed shapes presented on a touch screen with a two-alternative forced choice procedure. Ten baboons (OPEN + group) were trained with the open shapes serving as the positive stimulus (S+), and nine others (CLOSE + group) were trained with the closed shape serving as S+. The OPEN + group obtained higher discrimination performance than the CLOSE + group (Exp 1), but its scores declined when new line segments were added to the stimuli (Exp 2) and after smoothing the end points of the open shapes (Exp 3). The CLOSE + group was less affected by the above manipulations of local stimulus dimension, but its performance was disrupted when the collinearity end points was reduced (Exp 3). Use of a visual search task revealed that the search for an open shape among closed distractors was less attention demanding in baboons than the search for a closed shape among open ones (Exp 4). It is concluded that (1) end lines rather than closeness per se are perceptual primitives for the open versus closed discrimination in baboons, and (2) the relative emphasis on local or configural cues when processing contour closure depends on experiential factors in baboons and is thus subject to interindividual variations.