Emerging driver assistance systems, such as look-ahead cruise controllers for heavy duty vehicles, require high precision digital maps. This contribution presents a road grade estimation algorithm for fusion of GPS and vehicle real-time sensor data, with measurements from previous runs over the same road segment. The resulting road grade estimate is thus enhanced using measurements from additional traversals of known roads. Distributed data fusion is utilized to ensure that the storage requirement of known roads does not increase when additional measurements are processed. The implemented algorithm, which is based on extended Kalman filtering and smoothing, is described in detail. Experiments on a Scania test vehicle show the advantages and some of the challenges with the proposed approach.