In this article, a spatial gamut mapping algorithm (SGMA) which can preserve image details adaptively is proposed. First, the SGMA uses a guided filter to extract details that represent an image’s edges and textures from the luminance channel of the input image. The image details from outside the target gamut are then added back to the original image to address the detail loss that may be caused by gamut mapping. Then, based on the image details inside the target gamut obtained by statistics, the value of the “gamut mapping depth” is calculated dynamically for a specific value of the “in-gamut detail-preserving ratio.” Finally, in the detail-compensated image, the colors out of gamut are mapped into the target gamut uniformly according to the corresponding gamut mapping depth. The evaluation results based on paired-comparison experiments and image quality evaluation models show that the proposed SGMA has good performance on both preference and accuracy. According to the principle of the proposed SGMA, it has only one gamut mapping process for an original image, and there is no need to merge different frequency bands of the image. Therefore, the proposed SGMA not only improves the computational efficiency, but also fundamentally avoids the halo-artifacts caused by most SGMAs.