The number of components used in the compositional simulation of petroleum reservoirs affects both the accuracy and the speed of simulation results. A proprietary, fully automatic fluid characterization technology provides an efficient way to reduce the number of components of reservoir fluid samples used to simulate relevant laboratory tests within a predefined accuracy. This paper focuses on investigating how the number of components impacts the speed and accuracy of real world compositional reservoir simulations. The procedure starts with using PVT laboratory test data to characterize the reservoir fluid behavior under anticipated reservoir production mechanisms. The PVT laboratory test data are matched within a given accuracy by running the proprietary Fully Automatic Reservoir Fluid Characterization package, where the minimum number of components can be set by the user. The characterized EOS parameters are directly exported to a compositional simulator to perform the actual reservoir simulation using different numbers of components and two different simulation schemes, i.e., IMPES and Fully Implicit. Simulations are performed for a range of fluid types, from volatile oils to gas condensates. This paper presents a systematic investigation of how the number of components influences the accuracy of simulation results and the speed of the simulation. Three examples show that this method of reducing the number of components results in a significant increase in the compositional simulation speed while, at the same time, retaining the prediction accuracy.