Achieving a high rate of response to a mail survey is important from several viewpoints. Aside from any statistical considerations, it is obviously a waste of time and money to address envelopes and to mail questionnaires to people who do not return them. The Statistical aspects of the problem are even more serious. A low return usually means that the reported data are not representative of the universe the investigator is trying to sample. The various geographic areas may not be covered in their proper proportions and the few people who do respond from each locality may differ considerably from the average in that locality with respect to the item being estimated. Such differences tend to be predominantly in one direction. They lead to serious biases for which satisfactory adjustments cannot always be made. Even when the bias is kept under control by interviewing samples of nonrespondents, investigators find that it pays to have a high response to the mailed inquiry. The higher the response, the smaller the sample of interviewed nonrespondents needs to be to attain the desired level of statistical precision. Rates of response can be stimulated by several different devices. This paper considers one of these--the effects of personal visits to the individuals on a mailing list.