Use of mathematical modeling for determination of ultraviolet (UV) fluence in disinfection reactors requires accurate knowledge of the fluence rate distribution in a multiple lamp array. A method for measuring the fluence rate among a multiple lamp array was demonstrated using spherical actinometry. A matrix of four low-pressure UV lamps in air were investigated to evaluate the potential for shadowing and reflection to impact the fluence rate within and surrounding the lamp array. Two fluence rate distribution models were tested to determine the ability to predict the fluence rate distribution measured by the actinometers. Shadowing proved to attenuate UV light. Reflection from the lamp surface added 3-9% to the fluence rate, depending upon position in the reactor. These effects, as well as the fluence rate at various points in the lamp matrix were effectively modeled using RAD-LSI and UVCalc3D fluence rate distribution models. At fluence rates above 8mWcm(-2), the actinometry measured fluence rate was lower than the modeled rate, presumably from saturation of the actinometer solution at high fluence rates (close to the lamp). With multiple lamp reactors, the impact of shadowing can significantly affect fluence rate distribution and thus the level of microbial inactivation. If shadowing is not included in fluence rate distribution models, the fluence rate will be over predicted in the shadow zone of a neighboring lamp, falsely skewing model inactivation predictions.