In arid and semi-arid regions, trees are sparse or have difficulty surviving, and shrubs often occupy a dominant position in vegetation communities. The ecological functions and ecological value of shrubs are particularly worthy of attention in these regions. The sparse-elm grassland in Otindag Sandy Land is the climatic climax community; the trees are sparsely distributed, while the shrubs are densely distributed between the dunes or on the leeward slopes of the dunes. Shrubs have significant ecological functions in fixing sand dunes, improving soils, providing habitats, and increasing vegetation biodiversity. Shrub biomass accounts for a large proportion of the sandy land vegetation community. According to previous studies, shrubs are usually less important than trees, but the research on shrubs is not yet sufficient. The present shrub biomass simulation method is mostly adapted from the methods used for trees. However, the morphological characteristics of shrubs are obviously different from those of trees. A biomass prediction model specific for shrubs urgently needs to be developed. In this study, six dominant shrub species were investigated. Based on an allometric model, seven shrub measurements were used as predictors and compared to assess their abilities to predict shrub biomass. Among the seven measurements, a circular platform volume model, which has a structure close to the actual shape of a shrub, was designed as a new predictor. In this study, correlation coefficients and three model-fitting accuracy evaluation indices[including determination of the R² coefficient, significance index p-values, and standard error of the estimate (SEE)] were used to assess the prediction ability of the seven measurements. The results showed that:(1) Among the single measurements, the correlation coefficient between the crown diameter and biomass was highest, compared to those between height or ground diameter and biomass. (2) The correlation between the composite indices and shrub biomass was much stronger than that between the single measurements and biomass. In addition, the crown-related composite indices had stronger correlations with biomass than the ground diameter-related composite index did. This indicates that the crown diameter or crown-related composite indices may be better in predicting shrub biomass. (3) The circular platform volume model further improved the ability to predict shrub biomass. The results of correlation analysis and fitting analysis showed that the correlation between the circular platform volume and biomass was stronger and the fitting error was smaller compared with other prediction indicators. The correlation and fitting accuracy between circular platform volume and biomass were similar across different shrub species. This means that the circular platform volume model has better and more-stable prediction ability than other models for many shrub species, which implies that it is also more scalable. Therefore, we suggest that the circular platform volume is an ideal predictor when shrubs are adequately measured in the field. Otherwise, if the shrubs are not sufficiently measured, the crown diameter and crown-related composite indices are more suitable for the prediction of shrub biomass. In sum, this study established a biomass prediction model based on the circular platform volume of the six sandy shrub species, which provides a scientific basis for further study of the shrub carbon sink in sandy land and the ecological significance of shrubs in semi-arid and arid regions.