Objective <p>This study aimed to assess hippocampal microstructural alterations in patients diagnosed with temporal lobe epilepsy with hippocampal sclerosis (TLE-HS) and to evaluate the diagnostic utility of diffusion kurtosis imaging (DKI) in this population.</p> Methods <p>A total of 53 patients with TLE-HS and 53 healthy controls (HC) underwent brain magnetic resonance imaging, including standard sequences and DKI acquisition. Bilateral hippocampi were defined as regions of interest. DKI-derived parameters and the FLAIR signal intensity ratio (R_SI(FLAIR)) were compared between affected and unaffected hippocampi in patients with TLE-HS and with those of the HC group. The lateralization index of the bilateral hippocampi in TLE-HS was analyzed. Diagnostic performance was assessed using receiver operating characteristic curve analyses, and associations with clinical variables were explored.</p> Results <p>The hippocampus on the affected side in patients with TLE-HS demonstrated significantly increased mean diffusivity, axial diffusivity, and radial diffusivity, along with significantly decreased axial kurtosis (AK), mean kurtosis (MK), and radial kurtosis, when compared with both the unaffected hippocampus and the hippocampi of the HC group (all <i>p</i> &lt; 0.05). No significant differences in fractional anisotropy (FA) values were observed between the affected hippocampus and either the unaffected side or the HC group (<i>p</i> &gt; 0.05); however, FA values in the unaffected hippocampus differed significantly from those in of the HC group (<i>p</i> &lt; 0.05). In TLE patients, the lateralization index of DKI parameters and R_SI(FLAIR) in bilateral hippocampus was statistically significant except for FA. TLE-HS demonstrates exceptional discriminatory efficacy in differentiating the affected hippocampus from both the unaffected hippocampus and healthy controls. Among the parameters assessed, MD, AD, and AK emerged as the most effective single metrics, whereas FA exhibited limited efficacy and lacked statistical significance. The multi-parameter joint correction model significantly outperformed the single-parameter model, achieving complete differentiation of the affected side from healthy controls (AUC = 1.000). However, only a moderate distinction was observed between the unaffected side and the healthy controls. The combined model utilizing DTI and DKI provides substantial evaluative value for the epileptogenic side of the hippocampus. Among clinical variables, AK and MK values in the affected hippocampus were positively correlated with age at disease onset (<i>r</i> = 0.383, <i>P</i> = 0.005; <i>r</i> = 0.354, <i>p</i> = 0.009, respectively).</p> Conclusion <p>DKI-derived metrics provided sensitive indicators of hippocampal microstructural disruption in unilateral TLE-HS. These findings support the potential of DKI as a valuable adjunctive tool in the evaluation of hippocampal integrity in TLE-HS.</p>