There is a recognized need in the industry to improve the quality of our CFD (Computational Fluid Dynamics) processes. One part of that initiative is to measure the accuracy of the current processes and identify opportunities for improvement. This report documents the results of a disciplined calibration process that uses statistical analyses techniques to assess CFD quality. The process is applied to UH3D, a Navier-Stokes solver used at Ford to model vehicle front-end geometry and engine cooling systems. The study is focused on a Taurus under relatively ideal circumstances to address one of the major deliverables from the analytical process, i.e., what is the accuracy of the front-end cooling airflow predictions?To address this question, high quality isothermal experiments and calculations were conducted on twenty-three front-end configurations at four non-idle operating conditions. The radiator airflow data were statistically compared to quantify accuracy, to evaluate design trends, and to assess CFD quality.The study results indicate a good correlation between UH3D airflow predictions and experimental measurements. The average accuracy was -2.7% with a 95% confidence band of ±13.1%. Except for two configurations, the design trend information was excellent.The disciplined calibration process and statistical analysis techniques used in this investigation are generally applicable to many CFD quality initiatives. The technique preferred by the authors for assessing the reliability of the design predictions is the Spearman Rank Correlation method. It provides quantitative information about accuracy and makes it easy to identify opportunities to improve data quality. With much of the subjectivity removed, the quality of both CFD and experimental processes can be improved.