As the name suggests, design sensitivity was originally intended as a guide to the design of observational studies. This chapter considers three aspects of the design that affect the design sensitivity: (i) the block size, (ii) the exclusion of individuals who received small doses of treatment, and (iii) reducing the heterogeneity of treated-minus-control matched pair differences. Although references are given to theoretical results, the focus is on redesigning the study of alcohol and HDL cholesterol, noting the sensitivity to bias of alternative designs. Theoretical results differ by replacing actual data by a probability model with known properties and calculating the design sensitivity under various models.