The Center for Drug Evaluation and Research (CDER) is making increased use of observational studies in the evaluation of adverse outcomes associated with drug products. Several large initiatives including the MiniSentinel and Federal Partners programs are designed for this purpose. The evaluation may be part of routine active surveillance or a dedicated effort to address a hypothesized drug and adverse outcome association. These studies present several methodological challenges both for statistics and epidemiology, including confounder adjustment and sequential design. This talk will review the use of safety assessment with observational studies at CDER and present several recent examples. In particular, the use of propensity scores in studies will be highlighted and examined.