Objective <p>To evaluate the agreement of three different local thresholding algorithms, including Otsu’s local thresholding extension, Niblack, and Sauvola, in choroidal vascularity index (CVI) calculation using the ImageJ software.</p> Methods <p>The study used a cross-sectional design and involved healthy individuals with no refractive errors. All participants underwent imaging with macular enhanced-depth optical coherence tomography (EDI-OCT). The intraclass correlation coefficient (ICC) was calculated to evaluate the agreement between the three local thresholding algorithms in CVI measurement. Bland-Altman plots were also generated to assess the agreement and biases between pairs of algorithms. Two independent operators performed all image processing. The inter-rater and intra-rater agreements between the two operators were calculated.</p> Results <p>In this study, 46 eyes of 46 subjects participated. The mean ± standard deviation (SD) of the age was 24.37 ± 4.61 years (range: 18 to 35). The two-way mixed effects model showed no significant agreement between the three auto-local thresholding algorithms in CVI calculation (<i>P</i> = 0.992).</p> Conclusion <p>The study results emphasize the need for caution when employing automated local thresholding algorithms to measure CVI in OCT images. The lack of agreement in CVI measurements using Otsu’s, Niblack, and Sauvola algorithms may be explained by differences in their mathematical principles, sensitivity to contrast variations, and parameter tuning.</p>