BACKGROUND: To validate the clinical efficacy of non-invasive prenatal diagnosis (NIPD) for spinal muscular atrophy (SMA) in the first trimester and extend its applicability to families without probands. METHOD: From December 2020 to October 2024, 288 high-risk pregnancies were recruited prospectively, with 81 qualifying for NIPD after genetic counseling. Among the eligible cases, parent-based haplotypes were successfully constructed in 75 families (92.6%), while grandparent-based haplotype reconstruction was performed for the remaining 6 cases (7.4%) where proband samples were unavailable. Through targeted sequencing of the SMN1/SMN2 gene and flanking informative SNPs in maternal plasma, fetal haplotypes were inferred by analyzing dosage changes in cell-free DNA (cfDNA) using Bayes factor. All NIPD results were subsequently validated through invasive diagnostic procedures (chorionic villus sampling or amniocentesis). RESULTS: The haplotypes were successfully constructed in 81 families through parents or grandparents of the identified variant carriers. 76 families (93.8%) successfully obtained NIPD results, among which the earliest gestational week for successful NIPD was 7+3 weeks, with a minimum fetal fraction of 1.9%. 5 cases were classified 'no call' results due to pathogenic variant-adjacent recombination events (2/5), insufficient or unevenly distributed informative SNPs (2/5), and subthreshold fetal fraction (1/5). The average gestational age of NIPD blood drawing is 9 weeks. Validation test showed the NIPD results accuracy was 100%. CONCLUSION: This study demonstrates the clinical feasibility of grandparent-assisted haplotype construction for SMA families without probands and enables accurate early prenatal diagnosis of SMA in first-trimester pregnancies.