This article reports on a study to determine the nature and extent of
the problems discovered when cataloging records for a consortium of fi
ve libraries were merged to create an online public access catalog (OP
AC). Records in the host (University of Rhode Island) library's shelfl
ist were compared to records in both the Online Catalog Library Center
(OCLC) database and the OPAC. Problems were most acute in shared reco
rds, most commonly among those for reference materials and serials whe
re host library holdings were often missing. Even for records present,
call numbers were often missing, multi-volume records were often inco
mplete, and records for theses and dissertations were confusing and mi
sleading. A dirty database requiring some clean-up of past cataloging
sins had been expected. However, the magnitude of the problem and the
apparent overlaying of first-loaded records by subsequently-loaded rec
ords was totally unexpected by cataloging staff, thereby calling into
question the process by which the shared database was created.