Despite efforts to improve the quantity and quality of data for the assessment of resource status, there is a need for assessment methods that can be used in data-poor situations. This growing need is particularly driven by the latest revision to the Magnuson-Stevens Act (2007) which requires all federally managed fish stocks (those listed in various Fishery Management Plans) to have quotas (Annual Catch Limits) developed by 2011. Additionally, in the future we can expect increased demands for information on finer spatial scales to facilitate protection of stocks and substocks, and demands for information on more species in order to assess ecosystem structure and function. The concept of data-poor and data-rich stocks varies among the federally managed regions. At a minimum, a catalogue of features commonly encountered in data-poor situations would include:   incomplete data, possibly non-representative data, small sample sizes, limited number of variables for which there is information, uncalibrated data (e.g., unknown catchability and selectivity of fishing gear), unreliable data (e.g., unreported landings, unquantified discard mortality), and borrowing of parameter values or data from other species or fisheries. Methodological developments include modifying traditional estimators to deal with non-equilibrium situations and constructing general models that include data-rich and data-poor situations as special cases. This symposium focuses on the development, evaluation and application of assessment methods and models that are appropriate when the available data are extremely limited, with an emphasis on example cases.