Multiple-in multiple-out (MIMO) is a wireless communication technique where antenna arrays, at the receiver and transmitter, utilize signal multipath propagation to increase data throughput capacity. The unique benefits MIMO provides have over the last 20 years led to the steady increase in usage in both Wi-Fi and mobile networks. Predicting the performance of an antenna array designed for MIMO is more difficult than predicting the performance of a single antenna. This is due to the increased performance deriving from the processed combination of information from each antenna element. To determine the increased benefits that additional antenna elements can provide to a wireless system, the statistical correlation between the signals received from all antenna element needs to be evaluated. This correlation is expressed with the correlation coefficient ρ. The correlation coefficient may be estimated from the far field radiation pattern measured in an anechoic chamber, or measured from the statistically isotropic and homogeneous radiation environment provided by a reverberation chamber. However, Blanch, et al. 2003, proposed a much simpler method to estimate the correlation coefficient using a Vector Network Analyzer (VNA) to measure scattering parameters (S-parameters) while assuming perfect antenna efficiency. In 2005 Hallbjörner proposed a modified version of the estimation including the effect of antenna efficiency. This project aimed to measure and compare the results from the two types of chamber tests along with the two S-parameter based approximation methods mentioned. To accomplish this, three different antenna arrays, with four elements each with varying efficiency and mutual coupling, were designed and manufactured. The antenna arrays were then measured in an anechoic chamber, in a reverberation chamber, and had their S-parameters determined with a VNA. From the measurements it was found that the results from both types of chamber tests agree well, indicating that both tests are viable methods of signal correlation estimation. The S-parameter method proposed by Blanch was found to be inaccurate for the antennas tested, likely due to low radiation efficiencies. However, the approximation method proposed by Hallbjörner produced better results, but requires the efficiencies of the antennas which is generally not simple to determine. In conclusion it is found that S-parameter measurements, which are commonly used by the wireless industry, do not provide valid estimates of the MIMO performance of antenna arrays unless they are complemented with measurements of antenna efficiency.A special thanks to Bluetest, and especially John Kvarnstrand, for the reverberation chamber test opportunity and further guidance.