System and method for real-time, automatic, recognition of large time-scale driving patterns employs a statistical pattern recognition framework, implemented by means of feed-forward neural network utilizing models developed for recognizing, for example, four classes of driving environments, namely highway, main road, suburban traffic and city traffic, from vehicle performance data. A vehicle control application effects changes in vehicle performance aspects based on the recognized driving environment.