Accurate contour error calculation is crucial for improving the dynamic performance of machine tools, and the indirect detection method for test pieces serves as an important approach to evaluate this performance. This paper proposes a novel contour error calculation method for five-axis machine tools, integrating a flank-milled surface generation algorithm based on tool position information and a three-dimensional surface contour error calculation method utilizing normal curvature. First, the error-free flank-milled surface is generated using real-time tool position data. The theoretical surface is then divided into parameterized patches, and for each point on the actual flank-milled surface, the corresponding patch along the normal vector direction is identified. Finally, the contour error between the actual and theoretical flank-milled surfaces is calculated based on normal curvature. Comparative analysis with other contour error calculation methods demonstrates that the proposed method offers significant advantages in both accuracy and computational efficiency. This method provides reliable data for specimen detection, thereby contributing to enhanced dynamic performance of five-axis machine tools.