Landscape pattern is one of the important factors affecting plant diversity. However, the effect of landscape pattern in sandy areas remains unclear at various spatial scales. To investigate correlations between landscape pattern and plant diversity, as well as their scale dependence, the present study used field surveys and remote sensing technology to establish plant diversity indices and landscape metrics for the Hunshandak Sandland, northern China. Correlations among alpha diversity, beta diversity, and 33 landscape metrics were screened using bivariate correlation analysis at ten spatial scales (100, 200, 300, 400 500, 600, 700, 800, 900, and 1000m), and significant models were selected by stepwise regression analysis and verified by comparing fitted and field values. We found that landscape metrics influenced plant diversity in a scale-dependent manner The landscape pattern indices had a greater influence on Shannon diversity and β diversity than other diversity indices. The R² (0.138) of the Shannon diversity by stepwise regression was highest at the 100m scale, whereas the R² (0.2) of those of β diversity by stepwise regression was highest at the 900m scale. Comparison of the fitted and field values indicated that the Pielou, Shannon, and Whittaker indices were significant at the 500m scale, which confirmed that landscape pattern metrics significantly influence pant diversity at the 500m scale in sandy grassland. This demonstrates the importance of considering the scale effects of landscape pattern during plant diversity conservation and indicates that landscape pattern should be optimized and maintained in order to manage and preserve plant diversity in sandy grassland communities.