The explosion in the number and size of life science data resources, and the rapid growth in the variety and volume of laboratory data has been fueled by world-wide research activity and the emergence of new technologies. The m- eling, management and analysis of this data often requires a comprehensive - tegration of heterogeneous and typically semistructured data, distributed across many possibly data sources. Recent interoperability standards such as XML and WSDL solve some (easy) problems, but data and process integration often - main time-consuming and error-pone manual tasks. The di?culty of these tasks is compounded by the high degree of semantic heterogeneity across data sources, varying data quality, and other domain-speci?c application requirements. DILS 2005 was the 2nd International Workshop on Data Integration in the Life Sciences, following a successful ?rst DILS workshop, March 2004 in Leipzig, Germany. For a specialized workshop, the DILS 2005 call for papers created a largeinterest(over50abstractsandeventually42papersubmissions;anincrease ofover20%overDILS2004), outofwhichtheinternationalProgramCommittee selected 15 full papers, as well as 5 short papers, and 8 posters/demonstrations, which are all included in this volume. They cover a wide spectrum of theoretical and practical issues including scienti?c/clinical work?ows, ontologies, tools and systems, and integration techniques. DILS 2005 also featured keynotes by Dr. PeterBuneman, ProfessorattheSchoolofInformatics, UniversityofEdinburgh, and Dr. Shankar Subramaniam, Professor at the Department of Bioengineering andChemistry, UCSanDiego.Theprogramalsoincluded6invitedpresentations and reports on ongoing research activities in academia and industry and a panel organized by the AMIA Geomics Working Gr