Objective To develop a model using the recovery trajectory of margin reflex distance 1 (MRD1) to predict outcomes after conjoint fascial sheath (CFS) suspension for congenital ptosis. Methods Eighty patients (80 eyes) treated from January 2023 to January 2026 were retrospectively included and followed for 6 months. According to preoperative MRD1, 32 cases were moderate and 48 were severe; 64 had a good outcome and 16 a poor outcome. MRD1 was measured before surgery and at 1 week, 1, 3 and 6 months. A linear mixed model, logistic regression, ROC analysis, Bootstrap calibration and decision curve analysis were used. Results MRD1 increased at 1 week in both groups. The good-outcome group remained stable, whereas the poor-outcome group showed marked regression from 1 to 3 months (interaction P<0.001). Regression was greater in severe cases (P<0.001), with a higher poor-outcome rate (P=0.012). Preoperative MRD1, levator function, MRD1 ≥3.0 mm at 1 month and regression >0.5 mm from 1 to 3 months were independent predictors. The trajectory-enhanced model achieved an AUC of 0.891 (95%CI: 0.812-0.970), and the Bootstrap-corrected AUC was 0.874. Conclusion Persistent MRD1 regression from 1 to 3 months indicates insufficient fixation or tissue adhesion.