The conduct and interpretation of toxicology studies depends on an accurate measurement of the dose-response relationship. Critical to the study conduct is that the intended concentration of drug to be delivered be within 15% of the nominal dose. In GLP toxicology studies, the dosing formulations are evaluated after preparation to confirm accuracy, on at least one dosing day. What happens when a dose is substantially different from that intended during a pivotal GLP study? Herein, we present a root-cause decision tree developed from a case study in which the concentration of the dosing solutions varied from the nominal dose by as little as 4%, up to as much as 62%, the investigation conducted to determine the actual amount dosed on each dosing day, and the cause of the inaccurate doses.  Detailed documentation and reserve samples were critical to the investigation.  Hypotheses regarding the root cause were identified and systematically evaluated, including errors in test-article handling, dose preparation calculations, dose formulation analysis dilutions, contamination, degradation, and incorrect concentration of the bulk test article containers.  Using a decision tree and additional analytical testing, multiple potential process failures were ruled out, and it was determined that the bulk containers held test article that was unevenly distributed and not adequately mixed prior to aliquoting, and that this occurred on multiple occasions during work by multiple technicians.  The decision tree can be applied in any investigation of a dose preparation process failure, to determine the underlying cause and ensure that accurate dose levels are administered in future studies.  For this specific case, additional mixing steps and analytical testing were incorporated to ensure the use of homogeneous test article in preparing dose formulations for future studies that adhere to the 15% range about the nominal dose.