Objectives: Grain size analysis of clastic sediments is a crucial technical method for studying their depositional processes and environments. Existing methods such as sieving, laser diffraction, and thin-section image analysis suffer from differences in principles and limitations in applicability, hindering the accurate characterization of grain sizes across different size fractions and genetic types. To address this issue, this study proposes a stereomicroscopy image analysis method and systematically compares the applicability of various methods in clastic rock grain size analysis, with a focus on evaluating systematic biases and developing a correction model for thin-section measurement data. Methods: Four narrow-size-range standard samples of quartz sand (G1: 63–125 μm, G2: 125–250 μm, G3: 250–500 μm, G4: 500–2000 μm) were prepared for parallel experiments. The stereomicroscopy image analysis method, based on Greenough dual-light-path three-dimensional imaging, was introduced to avoid the two-dimensional "sectioning effect," and the Feret geometric mean diameter (Dgeo) was adopted as a morphology-compatible grain size characterization parameter. Results:(1) The average grain sizes measured by stereomicroscopy image analysis and laser diffraction are highly correlated, but the latter exhibits significant systematic underestimation (approximately 31%–37%); (2) Thin-section image analysis yields systematically smaller average grain sizes (32.28%–44.37%) due to the sectioning effect and particle irregularity. Numerical simulations quantitatively reveal that the errors originate from the sectioning effect (8.80%–21.84%) and morphological effect (22.53%–35.57%); (3) A correction model for thin-section image analysis average grain size is established based on the correction coefficient *k* (1.47–1.80); (4) Statistical analysis determines that the minimum sample size for stereomicroscopy image analysis is 300 particles, confirming its suitability for accurate analysis of coarse-grained (medium sand to fine gravel), micro-area (laminae), and trace (<1 g) samples. Conclusions: Stereomicroscopy image analysis demonstrates significant advantages in accuracy, measurement range, and sample requirements, making it an ideal method for characterizing irregularly shaped clastic particles, especially coarse-grained and trace samples. This study provides a scientific basis for selecting grain size analysis methods for different sample types and establishes a theoretical foundation for correcting thin-section grain size data and geological interpretation.