Background: Osteogenesis imperfecta (OI) is marked by clinical and genetic heterogeneity, and the genotype-phenotype correlation remains not very clear. We conducted a clinical and genetic study in a Chinese OI cohort to determine the spectra of phenotypes and pathogenic variants. Methods: In this study, 298 Chinese families were recruited from 2019 to 2024. Clinical phenotypes including fractures, short stature, skeletal deformities, blue sclera, dentinogenesis imperfecta, and hearing loss were recorded and analyzed. Next-generation sequencing combined with PCR-based techniques was used to detect candidate pathogenic variants. Variant pathogenicity was evaluated via conservation analysis, bioinformatics analysis, and functional studies at the cellular level. In this OI cohort, the spectra of pathogenic variants, clinical phenotypes, and genotype-phenotype correlations were analyzed. Results: Our OI cohort included 71 type I (23.83%), 122 type III (40.94%), 90 type IV (30.20%), and 15 type V (5.03%) probands. The cohort consisted of 196 children (65.77%) and 102 adults (34.23%). For the first time, phenotypic differences between different age groups were confirmed. In total, we identified 231 variants, including 47 novel pathogenic variants. Notable variants include two atypical splicing variants, one small deletion, two small duplications, one gross deletion, and one gross duplication. New genotype-phenotype correlations were observed: patients with SERPINF1 variants had the highest fracture frequency, followed by those with WNT1 variants, compared to patients with other gene variants. Conclusions: We performed the clinical and genetic analysis in a large Chinese OI cohort. The expanded spectra of genetic variants and clinical phenotypes were constructed by identifying 47 novel pathogenic variants and summarizing the skeletal and extra-skeletal manifestations. The current paper will provide important evidence for the precise diagnosis of the disease.