The compression sensing method reduces the times of sample inspection by sample mixing process, thus improving the sampling efficiency, reducing the sampling cost and shortening the sampling time. Methods Based on the characteristics of sparse data, the compression sensing method was applied to the food safety risk monitoring sample detection. The mixed samples were detected, and the original data was then reconstructed according to the corresponding algorithm with R statistical software. Results The detection values of 125 original samples were reconstructed using the compression sensing method, and the sum of squared errors was 3.782 652 × 10-29. There were 117 original samples lower than the detection limit were accurately reconstructed. The 8 samples which were higher than the detection limit were slightly larger than the real value, but the error is very small and can be neglected. Conclusion Compression sensing methods could reduce the testing number of samples by mixing samples and reconstruct the contaminant content of each original sample from several mixed samples.