The performance of 80 unselected patients with unilateral cerebral lesions (verified by CT scan) was compared with that of 34 control subjects on 6 "screening" tests for visual, auditory, tactile, kinaesthetic, motor and conceptual neglect. The performance of all 114 patients was tested comparably on the left and right sides, and (except for motor neglect) also centrally. Cut-off scores were determined so that performance inferior to 95% of the control range could be identified. The criterion for neglect was contralateral defect in the absence of ipsilateral and of central defect. With this procedure 30 patients were identified as showing neglect (5 visual, 6 auditory, 5 tactile and 4 motor--all 20 patients showing only one form of neglect; 10 with various combinations of neglect). In 27 of these 30 patients the lesion was found to be right-sided. The performance of various groups of patients with neglect was then compared with that of the 50 patients without neglect on 26 "evaluative" tests, designed to characterize the different varieties of neglect. For this comparison discriminant analysis was used. The outcome of two discriminant analyses instance a patient with visual and auditory neglect is not similar to a patient either with exclusively visual or with exclusively auditory neglect; (2) mixed, tactile and motor neglect are easy to discriminate from other kinds of neglect, whereas visual and auditory neglect are less easy to discriminate, particularly from patients without neglect; (3) the discriminant functions seem to reflect general spatial defect not confined to one side of space; differential spatial performance to the contralateral/ipsilateral sides; and manipulation. The findings are discussed in relation to previous interpretations of neglect; the defect is regarded as one of local attention.