This study examines the extent to which IT has differentially affected collaboration in the social sciences relative to the natural sciences. IT's impact on the social sciences may be larger because much research can be conducted virtually, while working in close proximity (in labs) may be more crucial to producing research in the natural sciences. To undertake the research, the authors match an explicit measure of institutional IT adoption (domain names, e.g. www.umsl.edu) with institutional data on all published papers indexed by ISI for 1,348 four-year colleges, universities and medical schools for the years 1991-2007. The publication data cover the social sciences, humanities, and natural sciences and narrower fields such as economics and biology. Three measures of co-authorship are examined: (1) average number of coauthors by institution; (2) percent of papers from an institution with one or more co-authors at another U.S. institution; and (3) percent of papers with one or more non-U.S. coauthors. The study describes collaboration patterns and then uses regression analysis to examine the impact of IT "exposure" on co-authorship. Preliminary results suggest: 1) dramatic growth in co-authorship within and across fields; and; 2) differential effects of IT by field.