This record is the input dataset used to develop the analytical equations described in the manuscript“Physics-Informed Analytical Models for Interpretable and Deployable Hydrogen Storage Prediction in MOFs” It is a frozen snapshot of the data used for training/validation in the symbolic-regression workflow; it is not a product of the article. The dataset aggregates crystallographic descriptors for MOFs together with GCMC-derived usable hydrogen capacities under 77 K, 100→5 bar, enabling exact reproduction of equation-discovery and benchmarking steps (SISSO, PySR, AI Feynman, gplearn). Scope Systems: Metal–Organic Frameworks (MOFs) Rows: 88,400 MOFs (after quality filters) Targets: Usable gravimetric capacity UCg (wt.%) and usable volumetric capacity UCv (g H₂ L⁻¹) at 77 K, 100→5 bar Descriptors (7): single-crystal density ρc (g cm⁻³), gravimetric surface area Sg (m² g⁻¹), volumetric surface area Sv (m² cm⁻³), pore volume Vp (cm³ g⁻¹), void fraction Fv (–), largest included sphere Di (Å), largest free sphere Df (Å) Contents MOF_SR_Training_Dataset.csv — table with columns: ρc, Sg, Sv, Vp, Fv, Di, Df, UCg, UCv (and any included identifiers) Units (key fields): UCg: wt.% · UCv: g H₂ L⁻¹ · Sg: m² g⁻¹ · Sv: m² cm⁻³ · Vp: cm³ g⁻¹ · ρc: g cm⁻³ · Di, Df: Å Provenance & Use Curated for physics-informed symbolic regression and comparative benchmarking. Suitable for reproducibility, baseline ML, and method comparisons. Please report any issues or clarifications to the contact below. How to Cite Dataset used to develop the analytical equations in “Physics-Informed Analytical Models for Interpretable and Deployable Hydrogen Storage Prediction in MOFs” (manuscript under review). Zenodo, DOI: 10.5281/zenodo.17108560.