We have used both quantitative and qualitative procedures to assess the validity of percutaneous penetration data obtained from the guinea-pig in vivo. Our results indicate that, under certain, not unusual, experimental conditions the results may be misleading and should be viewed with caution. Contamination of the experimental system, either by simple desquamation from the site of application, or trauma to the site such as rubbing or scratching, may greatly influence the results. We conclude that simple control experiments similar to those described should be conducted when using in vivo animal models for studies on percutaneous penetration, and particularly where the subject is not totally immobilized.