OBJECTIVE: Microcirculatory heterogeneity plays an essential role in the initiation and progression of glioblastoma (GBM). This study employs super-resolution ultrasound imaging to visualize the microcirculatory heterogeneity in GBM, with the objective of illustrating its predictive value in histological assessments. METHODS: This in vivo study explored the microvasculature in GBM models using 15 Sprague-Dawley rats, divided into three groups based on tumor growth stages (12, 18 and 24 d post-implantation). Ultrasound localization microscopy (ULM) was employed to assess microvascular morphology, hemodynamics and heterogeneity. Structural, functional and heterogeneity parameters at different tumor growth stages were quantified using Kruskal-Wallis H tests, or analysis of variance, followed by Bonferroni correction to characterize tumor progression. Linear correlations between these quantitative parameters and pathological indicators, including histological vascular density (VD-H), proliferation index and histological vascular maturity index (VMI-H), were evaluated. A stepwise linear regression model was constructed to assess the predictive performance in relation to histological parameters. RESULTS: Compared to histology, ULM enabled the earlier detection of tumor progression. The quantitative parameters derived from ULM provided a more comprehensive assessment than conventional metrics such as tumor size and immunohistochemistry. Multivariate analysis exhibited significant correlations among curvature, blood flow orientation variance (OV) and VD-H. Additionally, curvature, blood flow and OV demonstrated significant correlations with the proliferation index, while blood flow and fractal dimension showed significant associations with VMI-H. Heterogeneity parameters exhibited superior predictive power for certain histological features compared to microvascular morphology and functional perfusion. CONCLUSION: ULM provides a basis for early, non-invasive in vivo imaging and quantification of microvascular structures in rat GBM and demonstrates super-resolution predictive capability for histological parameters.