In Section 1.4 we stated a primary objective in the design of experiments to be the avoidance of bias, or systematic error. There are essentially two ways to reduce the possibility of bias. One is the use of randomization and the other the use in analysis of retrospective adjustments for perceived sources of bias. In this chapter we discuss randomization and retrospective adjustment in detail, concentrating to begin with on a simple experiment to compare two treatments T and C. Although bias removal is a primary objective of randomization, we discuss also its important role in giving estimates of errors of estimation.