Abstract Study question Could whole-exome sequencing (WES) be useful in clinical practice for primary infertile female involving oocyte or embryo defects and unexplained recurrent miscarriage (RM) patients? Summary answer Identifying key genes involved in oocyte and embryo development helps explain approximately 17% cases of reproductive failure. WES results will help optimize clinical therapeutic treatment. What is known already Dozens of genes have been reported to be the genetic causes of human infertility. Variants in these genes have specific effects on certain processes of human early reproduction and result in a spectrum of phenotypes, such as oocyte maturation arrest, fertilization defects, cleavage failure, early embryonic lethality, and recurrent miscarriage. Study design, size, duration This retrospective cohort study was conducted from July 2020 to August 2022. A total of 129 affected females were enrolled. Most of them were primary infertile women who suffered repeated failure of in vitro fertilization (≥2 cycles) due to abnormal development of oocytes or embryos (PI group, n = 125). Our cohort also comprised a small number of unexplained RM subjects without offspring (RM group, n = 4). All patients and their spouses had normal karyotype. Participants/materials, setting, methods The peripheral blood samples from all the participants and all of their available family members were obtained for DNA extraction. Genomic DNA samples from the probands were subjected to WES to identify candidate variants. Subsequently, the variants were validated by Sanger sequencing or qPCR. Familial co-segregation analyses were then carried out within the family members. The relationship between phenotype and genotype by clinical tests and the clinical records of the patients was studied. Main results and the role of chance Firstly, variants that are potentially relevant to the phenotypes of female infertility were identified in 17.05% of cases (n = 22), including TUBB8 in 14 cases, PATL2 in 2 cases, ZP2, ZP3, PANX1, TLE6, NLRP2 and NLRP7 in 1 case for each. A breakdown by phenotype revealed that we identified variants in 16.00% (n = 20) of PI group and 50% (n = 2) of RM group. Secondly, 16 TUBB8 variants were identified in 14 cases, of which 8 were novel, 3 were previously reported, and 5 were novel amino acid change occurring at the same position as another previously reported change. We found 3 TUBB8 variants (p.C211R, p.T232I, p.A342T) in 3 embryonic arrest families and 1 TUBB8 mutation (p.A102V) in a RM family, which were maternally inherited. Thirdly, 75 patients in PI group without phenotypic variant underwent ≥1 cycles of ovulation and/or frozen embryo transfer after WES, and 49.33% had achieved clinical pregnancy (n = 37). Limitations, reasons for caution The genomic DNA of parents were not available in all these 22 families, so we cannot determine whether the variant is de novo or inherited in some cases. Additionally, further functional studies should be performed to prove that these novel mutations affect protein function. Wider implications of the findings The present study expands the kinds of variants and phenotypic spectrum of gene mutations with regard to female reproductive failure. Genetic tests for causative genes involved in oocyte and embryo defects were recommended for primary infertile patients who suffered from these conditions and for unexplained RM patients. Trial registration number not applicable