To investigate the feasibility of constructing parental haplotypes based on short-read next-generation (NGS) and long-read third-generation sequencing (TGS) for non-invasive prenatal testing (NIPT) of thalassemia.Families at risk of having children with moderate to severe thalassemia were recruited for this study. Genomic DNA from both couples and sibling or grandparents of the fetus was first applied to construct haplotypes in couples using targeted NGS. For families where parental haplotypes could not be built using NGS, directly using TGS. NGS was performed on cell-free DNA in maternal peripheral blood to obtain information on fetal allele depth distribution, fetal fraction, etc. Haplotypes and thalassemia genotypes of fetuses were then deduced using the Viterbi decoding algorithm based on a hidden Markov model. Finally, the NIPT results were verified by invasive prenatal diagnosis (IPD). As a result, pedigree-based NGS successfully deduced the thalassemia genotype of the fetus in 93.3%(28/30). However, 2 families had failed because both sibling or grandparents of the fetus and couples were heterozygous for point mutations in the same thalassemia gene. In these 2 families, TGS directly inferred parental haplotypes and deduced fetuses genotypes. This combined strategy resulted in the prediction of fetal genotypes in all 30 families, and its coincidence rate with IPD results reached 100%. In brief, Pedigree-based NGS is enough to construct haplotype and decuce genotypes in most fetuses. TGS is helpful in certain families in which NGS failed. Therefore, comprehensive application of haplotype analysis based on NGS and TGS data is an effective strategy for NIPT in thalassemia.