Technology evolution forecasting based on historical data processing is a useful tool for quantitative analysis in technology planning and roadmapping. While previous efforts focused mainly on one-dimensional forecasting, real technical systems require the evaluation of multiple and conflicting figures of merit at the same time, such as cost and performance. This paper presents a methodology for technology forecasting based on Pareto (efficient) frontier estimation algorithms and multiple regressions in presence of at least two conflicting figures of merits. A tool was developed on the basis of the approach presented in this paper. The methodology is illustrated with a case study from the automotive industry. The paper also shows the validation of the methodology and the estimation of the forecast accuracy adopting a backward testing procedure.