The number and volume of remote sensing data and its derived products, which are regarded as typical “big data”, have grown exponentially. How to assess the quality of these big remote sensing products become a challenge. As an importance technique, spatial sampling is regarded to be necessary for the quality assessment of remote sensing derived products. This paper proposes an approach of multiple stratified spatial sampling for assessing the remote sensing products, with the aim of resolving the issue of the quality inspection of big remote sensing products. The proposed method improves the sampling accuracy without increasing the sampling size, and the whole procedure is repeatable and easily adopted for the quality inspection of remote sensing derived products.