Gurbantunggut Desert is a famous temperate desert with plant diversity is more special than the same latitude desert in the world. It is an important desert from the viewpoint of drought resistance, heat resistance, and saline plant germplasm resources. Samples were collected from 70 permanent plots (100 m × 10 m) in the desert, and the total number of different types of seeds and spatial patterns were investigated. Variations in seed mass and seed projected area among 87 species in different taxa were analyzed by measuring the 100-seed weight. The results were as follows: 1) The seed mass of plants can be divided into five classes: A (0.1-1 mg), B (1-10 mg), C (10-100 mg), D (100-1000 mg), and E (1000-5000 mg). These classes comprised 1.1%, 3.5%, 34.5%, 47.1%, and 13.8% of all species, respectively. Class D seeds were the most commonly examined species. The species were similar to those found in Mediterranean deserts, such as those in the northern areas of the Tel Aviv Poleg Nature Reserve. Most possess seeds of medium mass. Because the genera move from the Mediterranean region through western and central Asia, the seed mass of the desert plants may be determined by phylogenetic evolution. 2) There was a significant difference in seed size among the family, maximum family of 100 grain weight is Polygonaceae (29315 mg), the smallest is Redanko (0.3 mg). There was significant difference in seed size among genera such as Chenopodiaceae, Compositae, and Boraginaceae. 3) The 87 species included six life forms and covered significant differences in seed mass as follows: shrub > ephemeral plant > perennial herb > ephemeroid > subshrub > annual herb. A great deal of shrub seed mass belonged to type E (1000-5000 mg). 4) There was no significant correlation between seed size and species relative distribution frequency. However, removed the environmental specialization species, which relative distribution frequency < 1.4%. A negative correlation was observed between the exponential growth model of seed size and species distribution relative frequency. 5) The total number of species with type B, C, and D seeds increased with significant increase in altitude, but decreased with significant increase in latitude. It also increased with significant increase in longitude and precipitation. This was also true of the total number of seeds in the sample area. However, the opposite of this was true for type E seeds. This indicates that more precipitation gradually offered an advantage to plants with type B, C, and D seeds in the area but disadvantage to plants with type E. In future, efforts should be taken to protect the shrub belonging to type E in the Gurbantunggut Desert. The results of this study may facilitate the understanding of seed mass in the life history strategy of plants in arid desert areas and lay the foundation for further studies of plant biology in the Gurbantunggut Desert.