Determining the best methods of assessment for a library instruction program in a large research university can be a challenging task. Albert R. Mann Library at Cornell University Library has pilot tested three methods of formative and summative assessment for its library instruction programattitudinal, outcomes-based, and gap-measure-and determined not only key areas of improvement for the program, but also the benefits and drawbacks of each method of assessment. Attitudinal assessment has guided program improvement in areas of marketing and user satisfaction but does not provide the measurement of learning that outcomes-based assessment covers. The latter can be difficult to achieve in single-session, short-term instruction, while gap-measure assessment can provide a more nuanced view of both patron and instructor attitudes toward learning outcomes, if not actual data on achievement on the objectives themselves. The authors have determined that a combination of these three different types of assessment can address the shortcomings of a single method alone and provide a better measure of the program as a whole.Keywords: information literacy, assessment, evaluation, college and university libraries Assessment of library instruction is a growing topic of concern for library instructors as evidenced by the 246 articles found in Library Literature on evaluation of bibliographic instruction (more in the last two years than in the previous six combined). A number of assessment models are available for library instruction; however, with the wide range of methods and their varying utility depending on the format and type of instruction being assessed, choosing and implementing models can be difficult. It is our contention that to present a well-rounded picture of an instruction program, more than one method of assessment is necessary. Over the last two years, the staff at Albert R. Mann Library at Cornell University Library (CUL) has pilot-tested a three-pronged approach to assessment. The first assessment was an evaluation of the entire CUL instruction program using focus groups and a Web-based statistical survey; the second outcomes-based assessment, focused on a single courserelated instruction session at Mann Library, consisted of a pre-test survey and examination of the results of a post-instruction tutorial assignment; and a third assessment featured a gap-measure analysis of instructors' and students' perceptions of the importance of the learning objectives in Mann Library's open workshop series. Below we present our findings and their effect on our instruction program, address the benefits and limitations of these various types of assessment, and show that a multi-method model of assessment, both formative and summative, is key to a clear picture of the various aspects of our instruction program.