The Manchester Child Attachment Story Task (MCAST) is a representational technique for assessing attachment patterns of young school-age children. We have developed a computerised version (the CMCAST) in which story stems are represented on the computer by the movement of simple screen ‘dolls’. This paper reports on a preliminary validation study of the CMCAST method against the MCAST. Fifty-five children completed the MCAST and CMCAST six weeks apart in random order. It proved possible to rate the CMCAST if a simplified form of the MCAST coding system was used. Inter-rater reliability was achieved for both versions (kappa = 0.93 for MCAST and kappa = 0.91 for CMCAST). Agreement between the MCAST and CMCAST ratings of attachment security was kappa = 0.67. Costs for the MCAST and CMCAST were comparable. A school-based feasibility study of 86 children suggested that the CMCAST was acceptable and could be administered with up to five children simultaneously. This preliminary study suggests that the CMCAST can reliably reproduce a simplified form of MCAST coding. The computer format may be well adapted to some uses such as screening for large-scale epidemiological research.