The case for test marketing is now widely accepted, and most of us here have been involved in carrying out research for test markets. The risks associated with modern mass marketing are so considerable that manufacturers gladly avail themselves of any research tool which will reduce these risks. This is the object of test marketing. The theory is simple - we replicate in miniature the marketing situation (new product, new pack etc.), we observe what happens, and project the answers to national proportions. The object of this paper is to illustrate the practical difficulties involved in applying this theory.