IMPORTANCE: Maternal hemoglobin is a clinically accessible biomarker of preterm birth (PTB) risk, but abnormal thresholds are inconsistent and do not account for variation by gestational age or altitude. OBJECTIVES: To evaluate the association between maternal hemoglobin concentration and PTB across gestational ages and altitudes, including PTB subtype, and to define gestational age-specific, altitude-adapted hemoglobin thresholds and assessment windows. DESIGN, SETTING, AND PARTICIPANTS: This multicenter, retrospective cohort study used a multisource database integrating electronic medical records, preconception examination, and geospatial data from 5 medical centers in Yunnan, China, a region with a marked altitudinal gradient, and consecutively included singleton pregnancies delivered between January 1, 2017, and July 31, 2025. EXPOSURE: Prepregnancy and prenatal hemoglobin values were assessed. Prenatal measurements were grouped into 2-week intervals (target gestational week ±7 days), with 36 weeks analyzed separately. MAIN OUTCOMES AND MEASURES: Preterm birth was operationally defined as live birth from 28 weeks 0 days through 36 weeks 6 days of gestation. RESULTS: Of 141 943 singleton live births among women (mean [SD] maternal age, 29.7 [4.6] years), 12 607 (8.88%) were PTBs. Maternal hemoglobin measurement showed a gestational age-dependent U-shaped association with PTB, and thresholds varied across gestation. Compared with the intermediate group, a low hemoglobin concentration measured at 13 weeks ( 13.50 g/dL; AOR, 1.28; 95% CI, 1.04-1.55). At high altitude, thresholds were generally higher, and a low hemoglobin concentration from 21 weeks was persistently associated with PTB. Subtype analyses showed earlier low hemoglobin concentration associations for medically indicated PTB and middle- to late-pregnancy high hemoglobin concentration associations for both subtypes. CONCLUSIONS AND RELEVANCE: In this cohort study, hemoglobin thresholds associated with PTB varied across gestation and were generally higher at high altitude; low- and high-hemoglobin assessment windows differed and varied by altitude. These findings may help refine gestational age- and altitude-specific interpretation of routine hemoglobin measurements and PTB risk assessment, pending external validation.