Purpose <p>To evaluate the diagnostic efficacy of time-dependent diffusion MRI (t_d-dMRI)-based cell size imaging to distinguish gliomas with different isocitrate dehydrogenase (IDH) genotypes from lung cancer or breast cancer brain metastases (BrMs).</p> Methods <p>86 patients (29 males; mean age, 52.52 ± 12.12 years), including 21 patients with isocitrate dehydrogenase (IDH)-wildtype gliomas, 10 with IDH-mutant gliomas, 33 with breast cancer BrMs, and 22 with non‑small cell lung cancer (NSCLC) BrMs, were prospectively recruited for t_d-dMRI examinations. Microstructural parameters of the contrast-enhancing tumor and peritumoral edema, including mean cell diameter (C_d), volume fraction of the intracellular space (V_in), extracellular diffusivity, and cellularity, were estimated from t_d-dMRI using the IMPULSED model and compared across different tumors. Diagnostic performance was assessed using the area under the receiver operating characteristic curve (AUC). The microstructural parameters were validated with pathologic measurements.</p> Results <p>The contrast-enhancing tumor regions of NSCLC BrMs had significantly higher C_d and V_in values compared with those of IDH-wildtype gliomas, IDH-mutant gliomas, and breast cancer BrMs (<i>P</i> &lt; 0.001 - <i>P</i> = 0.002). C_d of the contrast-enhancing tumor performed well in distinguishing NSCLC BrMs from IDH-wildtype gliomas (AUC = 0.864), from IDH-mutant gliomas (AUC = 0.923), and from breast cancer BrMs (AUC = 0.904). IDH-wildtype and IDH-mutant gliomas had no significant differences in all parameters (<i>P</i> &gt; 0.05). No significant differences were found in all parameters in the peritumoral edema region (<i>P</i> &gt; 0.05). C_d correlated well with pathologically average cell diameter (<i>P</i> = 0.013, <i>r</i> = 0.770) and total cell area (<i>P</i> = 0.039, <i>r</i> = 0.673) of brain tumors.</p> Conclusion <p>t_d-dMRI-based microstructural imaging can be used to distinguish NSCLC BrMs from gliomas and from breast cancer BrMs.</p>