Reconstructing the spectral reflectence of the object surface from RGB signals of digital camera is one of the important studies of spectral color managament. A new algorithm based on back propagation (BP) neural network and principal component analysis (PCA) is proposed to realize the spectral reflectence reconstruction of color atlas. The optimal structure of BP neural network and the number of principal components are studied in the spectral reflectence reconstruction experiments of three color atlases and the accuracy of the algorithm is also testified. The experimental results show that the new algorithm of appropriate BP neural network combined with PCA is satisfied to accurately reconstruct the spectral reflectence of the same kind of color atlas.