Recent biotechnological advances have led to accelerated rates of development and improvement for a wide variety of vaccines. The need to optimize the evaluation of vaccine safety, immunogenicity, and efficacy for specified populations is becoming increasingly important. This study will extend, adapt, and apply modern statistical methods for handling detailed covariate information to the evaluation of vaccines at all phases of development. Special attention will be given to problems of correlation induced by longitudinal measurements or clustered samples. For Phase I and Phase II trials, asymmetric least squares and bootstrap techniques for percentile and nonlinear regression analyses will be investigated and extended for the analysis of repeated measures of immune system responses. Public domain software for these analyses will be developed. For prospective Phase III trials, the implementation of a variety of correlated data techniques when there has been randomization by cluster will be investigated. These methods rely, in general, on large sample asymptotic justifications; this study will develop techniques with improved second-order asymptotic properties. For Phase IV observational and case-referent studies, extensions of generalized estimating approaches to accounting for the correlation in clustered data will be studied. New methods of analyzing case-exposure data will be developed. Extensions of transmission models for handling dependence will be made and compared to current methods. Finally, propensity score methods will be applied to non-randomized vaccine studies and their properties evaluated. The results of this study will help shorten the vaccine development and licensure process by earlier needs identification and problem detection. In addition, the development of improved statistical techniques will yield more efficient study designs and precise quantification of relationships, providing savings in research costs.