Optimization-based damage identification in continuous rigid frame (CRF) bridges faces challenges. As the degrees of freedom of the structure increase, the complexity of the objective functions increases significantly, making convergence more difficult to achieve. This study introduces a stiffness separation method for damage identification in CRF bridges. This method decomposes the overall stiffness matrix of the bridge into the required stiffness submatrices, which greatly simplifies the objective function for identifying structural damage. By dividing the stiffness matrix into smaller stiffness submatrices, the proposed method reduces computational complexity and improves damage detection efficiency. Two CRF bridges are presented to verify the effectiveness of the proposed method.