This paper studies the flow and temperature patterns in an overhead diffuser based data center. In-situ measurements of the data center were carried out to validate a mathematic model for predicting the effect of different air distribution systems. With the measured data of temperatures and airflow velocities, the mathematic model is constructed using a commercial Computational Fluid Dynamics (CFD) software and experimental data to present a comparison between test results and numerical simulations. The area of the data center is 311 square meters and the heat load of the equipment is 320~360 watt per square meter. In-situ temperatures and humidity of the data center were measured with an Automatic Temperature and Humidity measuring instrument, whose error is ±0.5 °C. The discrepancy of the temperature and velocity between the numerical and experimental results were within ±2.3 °C and ±1.8 m/s, respectively. In addition, analysis shows that changing the volume flow rate of the cold air delivered to some diffusers can optimize the temperature field and thereby save the energy.