PurposeScience policy and practice for open access (OA) books is a rapidly evolving area in the scholarly domain. However, there is much that remains unknown, including how many OA books there are and to what degree they are included in preservation coverage. The purpose of this study is to contribute towards filling this knowledge gap in order to advance both research and practice in the domain of OA books.Design/methodology/approachThis study utilized open bibliometric data sources to aggregate a harmonized dataset of metadata records for OA books (data sources: the Directory of Open Access Books, OpenAIRE, OpenAlex, Scielo Books, The Lens, and WorldCat). This dataset was then cross-matched based on unique identifiers and book titles to openly available content listings of trusted preservation services (data sources: Cariniana Network, CLOCKSS, Global LOCKSS Network, and Portico). The web domains of the OA books were determined by querying the web addresses or digital object identifiers provided in the metadata of the bibliometric database entries.FindingsIn total, 396,995 unique records were identified from the OA book bibliometric sources, of which 19% were found to be included in at least one of the preservation services. The results suggest reason for concern for the long tail of OA books distributed at thousands of different web domains as these include volatile cloud storage or sometimes no longer contained the files at all.Research limitations/implicationsData quality issues, varying definitions of OA across services and inconsistent implementation of unique identifiers were discovered as key challenges. The study includes recommendations for publishers, libraries, data providers and preservation services for improving monitoring and practices for OA book preservation.Originality/valueThis study provides methodological and empirical findings for advancing the practices of OA book publishing, preservation and research.