It is apparent from studying recent articles on pharmacokinetics that a number of misunderstandings exist, both about the design of experiments and the analysis of results. The purpose of this paper is to outline many of the common pitfalls associated with the design of experiments and also the limitations upon the analysis of results. The paper describes mathematical, laboratory and clinical aspects which must be examined in designing a protocol for pharmacokinetic experiments. Simulated data is presented to demonstrate the dangers of using standard computer programs for parameter estimation. Even when convergence is obtained the answers may be dependent on the method employed. A mathematical model is of little use unless a reasonable amount of good, accurate data is obtained.