A comparison is made of the influence of two computer-controlled traffic signal timing plans over a given route: the existing plan based on SIGRID (SIgnal GRId Design program) and the proposed one, TRANSYT (TRAffic Network StudY Tool). Under the TRANSYT timing plan, vehicles encountered fewer stops, saved time and used a slightly smaller amount of fuel than under the existing timing plan. Vehicle fuel consumption was calculated with a computer model of a vehicle which used velocity profiles obtained from an instrumented "floating" car. Single and multiple linear regression analyses were used to determine the relationship between the fuel consumption and relatively easy-to-measure and statistically stable quantities such as trip time, number of stops and delay time. It was found that fuel consumption could be expressed adequately as a linear combination of trip time, number of stops and delay time. Using only two independent variables showed a combination of delay time and stops to be as good as a combination of travel time and stops. When restricted to a single independent variable, any one of these could be used for predicting fuel consumption.