LEDs (light-emitting diodes) have been widely adopted for use within traffic signals, recently replacing incandescent bulbs. LEDs degrade slowly – unlike incandescent bulbs that fail catastrophically. When the luminous intensity of LEDs falls below a pre-specified threshold, they pose danger to traffic. The long-term performance and degradation rates of LEDs have not been thoroughly studied in order to gain an understanding of their useful lives and appropriate replacement schedules. There exist many stochastic factors that affect LED degradation rates making their analysis complicated. This article provides a statistical methodology based on ordinary least-squares regression for measuring the useful life and the degradation rate of an LED signal, and presents details from a field study conducted in Missouri, U.S. Our results indicated that signal type, color, and manufacturer affect degradation, and therefore useful life should be calculated for each subgroup of LED traffic signals separately. Results of this research provide a much needed methodology for engineering managers in departments of transportation and local communities for replacing LEDs.