The purpose of this study was to refine the instrument, Nurses' Computer Attitudes Inventory (NCATT), which was developed to measure nurses' attitudes toward computers in an Australian setting. The study was designed to test the reliability and validity (concurrent and discriminant) of the instrument. The NCATT was administered to 170 subjects: 71 first-year nursing students and 99 nurses employed in a local hospital. On the basis of factor and item analysis the 40-item NCATT was reduced to 22 items. Three factors for the revised NCATT were identified: (1) Computers and Patient Care, (2) Computer Anxiety, and (3) Patient Confidentiality and Computers. These three factors accounted for 90.0% of the variance (factor 1, 56.6%; factor 2, 24.8%; and factor 3, 8.6%). The factors demonstrated good internal consistency with the Cronbach alpha coefficients for each factor ranging from 0.72 to 0.90. The revised NCATT provided evidence of concurrent validity on the student sample when related to Dambrodt's scale for Computer Attitudes. Some evidence of discriminant validity was demonstrated as the internal consistency reliabilities were much higher in all factors than their intercorrelations. The authors propose that the revised NCATT is a practical instrument that is useful to assess nurses' attitudes before computer implementation and training. © 1996, Lippincott-Raven Publishers.