Ghost charts, sometimes referred to as shadow charts, are duplicate medical records. Governance documents in several countries suggest that ghost charts present a risk to patient safety, to the extent that they contain information which may not appear in an official hospital record. Although most would agree ghost charts should not exist, their existence is widespread. This paper reports on an in depth multi-method qualitative study of ghost charts undertaken in two ambulatory care settings in a Canadian hospital. The study was undertaken in order to inform the design and implementation of a clinical information system which it is hoped will eliminate the need for duplicate charts. Our research demonstrated that ghost charts filled a variety of needs only some of which are typically accounted for in electronic record design. We suggest that if the functions ghost charts fill are not addressed, their existence will persist. This work is significant in that few studies of ghost charts have been undertaken, and in the in-depth understanding it contributes to design requirements for electronic record systems.