The purpose of this study was to assess the diagnostic sensitivity of tasks employing feature and conjunction visual searches in stroke patients with unilateral spatial neglect (USN). Seventy-two stroke patients (right/left hemispheric damage with/without USN) and 39 healthy controls participated in the study. Hit rate and reaction time measures of feature and conjunction searches were tested using a newly developed computerised programme for the assessment of visual spatial attention (VISSTA). In addition, subjects received a set of diagnostic paper-and-pencil tests, and were also assessed for the impact of neglect on activities of daily living. Results indicated that the computerised test clearly differentiated between stroke patients and healthy controls, and between the different patient groups. USN patients showed significant contralesional disadvantage in both feature and conjunction visual search tasks. It is proposed that computerised assessment of visual search capacity is a useful and sensitive adjunct to standard paper-and-pencil tests of USN, with the advantage of testing responses based on attention shifts under a time constraint. The learning effects that limit the usefulness of paper-and-pencil tests in longitudinal studies are less likely to affect a computerised test, making it more suitable for monitoring treatment-induced or natural recovery by way of repeated testing.