Research about service operation costs in the rail sector has usually focused on freight, high-speed, or national passenger rail, but has seldom included the study of the cost of urban rail (metro) rapid transit. This study analyzed the determinants of train service costs for a panel of 24 metro systems worldwide. The study used econometric modeling to assess the relative weight of each factor. Wages and electricity prices and consumption were found to have statistically significant elasticities and evidence of potential substitution effects between factors. Other factors, such as driver productivity, network length, percentage of rolling stock with air conditioning, and rolling stock age, also showed statistically significant elasticities. The study found evidence of strong returns to density and returns to scale in the provision of train service outputs (for example, car kilometers, passenger journeys, and train hours).