Take soil organic matter(SOM) as an example, this study discussed the spatial variability and rea-sonable sampling number for soil organic matter. Sampling schemes with different sample intensities were simulated by extracting data from a soil nutrient database of intensive samples. The results showed that sam-piing density had no significant influence on spatial correlation and model fitting of soil nutrient on a certain scale. It is possible to reduce the number of sample points to satisfy the need of the research regarding spatial variability of soil nutrients. Taking spatial prediction accuracy into consideration, more than 400 samples should be the suitable number to evaluate the spatial variability of SOM on county scale.