This study presents a novel hybrid model based on a two-dimensional (2-D) site-specific model and a statistical model to investigate space diversity in indoor environments. The statistical model describes the field scattered by rough surface boundaries and randomly positioned scatterers which may be important when one or both of the receiving and transmitting antennas are close to the boundaries of the scatterers. Comparing the computed spatial correlation with the measured one with the transmitting frequency at 2.44 GHz at many different sites demonstrates the effectiveness of the hybrid model. In addition to accurately predicting field strength the hybrid model can quantify the relative mean contribution of diffused scattering in an indoor environment with a factor r. The factor is equal to the ratio of ensemble average of a randomly scattered envelope to the spatially averaged envelope. Its optimum value is in a narrow range from 0.3 to 0.5 when the intensity fluctuation at the measurement sites is in or close to saturation regions where a large number of micromultipaths are generated. This finding confirms the relative ease in applying the hybrid model. Experimental results suggest that choosing r = 0.4 in a saturation allows the hybrid model to yield a satisfactory performance as confirmed by a blind test. Moreover good diversity gains can be obtained with an antenna spacing greater than or equal to one wavelength. Our results further demonstrate that the diversity gain of horizontally spaced antennas exceeds that of vertically spaced antennas.