This paper attempts to indicate the relevance of Broadbent's recent views on the methods by which the systematic selection of information takes place, for the study of attention deficits in schizophrenia. A two-stage model of attention is briefly described, the distinction between ‘filtering’ (stimulus set) and ‘pigeon-holing’ (response set) being emphasized. In an attempt to clarify the nature of the schizophrenics attention defect, various experiments are reviewed and interpreted within Broadbent's theoretical framework. Studies indicate inefficiencies in both the ‘filtering’ and ‘pigeon-holing’ mechanisms. The need for experiments specifically designed to identify the locus of the schizophrenic attention defect, in various subgroups of schizophrenics, is indicated.