The invention discloses a roller compacted concrete compaction degree evaluation method based on a GA-BP network. The method comprises the following steps: selecting the moisture content, the rollinglayer surface stress transverse wave velocity, the rolling material grading factor and the rubber sand ratio of a rolling material at each measuring point of a construction site as input index parameters of a real-time evaluation model; Determining a neural network structure of the compaction degree real-time evaluation model; Optimizing an initial weight value and a threshold value by utilizing agenetic algorithm; Substituting the determined initial weight value and the threshold value into a BP neural network for fine tuning to establish an optimal neural network model; And performing real-time evaluation to obtain a compaction degree value. According to the method, BP neural network is adopted on the basis of the moisture content of the roller compaction material before concrete rolling compaction, the surface stress wave velocity of a measuring point after rolling compaction is completed and corresponding concrete rolling compaction material grading characteristic parameters; TheGA-BP neural network constructs a compaction degree prediction model, predicts and evaluates concrete compaction degree indexes at real-time measurement points of rolling compaction, and can provide amethod for reliably judging the accuracy of the real-time compaction degree in the on-site construction process.