Optimality of inference procedures as well as new methodology in statistics are to be studied. The problem areas include (1) unbiasedness and monotonicity of power functions of tests; (2) inference in curved exponential families; (3) analysis of incomplete block designs; (4) variable selection techniques; (5) extensions of James-Stein results to non-normal cases; (6) extension of James-Stein results to non-quadratic losses.