The evaluation of reference service is complex and subjective, because reference service itself is complex and subjective. Reference service is more easily evaluated if its facets are judged against corresponding, known criteria. The core facet of reference service, for an automated library, is its online public access catalog (OPAC). Although the library literature contains numerous papers on the functional and performance evaluation of OPACs, as well as on the evaluation of many facets of reference service, it presents little assistance for the evaluation of OPACs as the central facet of reference service. In order to alleviate this lack, this paper evaluates OPACs as if they were any other reference tool, judging them against Norman D. Stevens’ classic eighteen criteria for the evaluation of reference books. A selective bibliography of works on both OPAC and reference book evaluation is included.