Nighttime crashes are overrepresented on United States and state roadway systems. Nighttime safety at signalized intersections in urban areas receives more attention because of frequent and serious vehicle-to-vehicle, vehicle-to-pedestrian, and vehicle-to-bicyclist traffic conflicts at night. Street illumination has proved to be an effective countermeasure to increase visibility at intersections and to reduce nighttime crashes. However, the safety effects of illuminance at urban signalized intersections are not well documented. This study investigated the impacts of illuminance on nighttime crash occurrence and nighttime crash injury severity at urban signalized intersections. Illuminance data and crash data for a sample of 91 signalized intersections were collected from the Tampa Bay region in Florida. The negative binomial model and the probit model were developed to examine the safety effects of intersection illuminance in terms of crash frequency and the risk of fatality and severe injury, respectively. The models showed that an increase in intersection illuminance from low (<0.2 fc) to medium (≥0.2 fc and <1.1 fc) could reduce nighttime crash frequency and night-to-day crash ratios by approximately 50%. When illuminance was kept at 0.9 fc or higher, the risk of fatality and severe injury decreased significantly, especially in crashes that involved pedestrians and bicycles, head-on crashes, and angle crashes.