Background: Schizophrenia (SCZ) is a complex and severe mental illness. There is a lack of effective biomarkers for SCZ diagnosis. The aim of this study was to explore the possibility of using serum peptides for the diagnosis of SCZ as well as analyze the association of variants in genes coding for these peptides and SCZ. Methods: After bead-based fractionation, the matrix-assisted laser desorption ionization/time-of-flight mass spectrometry technique was used to identify peptides that showed different expressions between 166 SCZ patients and 201 healthy controls. Differentially expressed peptides were verified in a second set of samples (81 SCZ patients and 103 healthy controls). The association of SCZ and three tagSNPs selected in genes coding for differentially expressed peptides was performed in 1,126 SCZ patients and 1,168 controls. Results: The expression level of peptides with m/z 1,945.07 was significant lower in SCZ patients than in healthy controls (P < 0.000001). The peptide with m/z 1,945.07 was confirmed to be a fragment of Kininogen-1. In the verification tests, Kininogen-1 had a sensitivity of 95.1% and a specificity of 97.1% in SCZ prediction. Among the three tagSNPs (rs13037490, rs2983639, rs2983640) selected in the Cystatin 9 gene (CST9) which encodes peptides including Kininogen-1, tagSNP rs2983640 had its genotype distributions significantly different between SCZ patients and controls under different genetic models (P < 0.05). Haplotypes CG (rs2983639–rs2983640) and TCG (rs13037490–rs2983639–rs2983640) were significantly associated with SCZ (CG: OR ¼ 1.21, 95% CI [1.02–1.44], P ¼ 0.032; TCG: OR ¼ 24.85, 95% CI [5.98–103.17], P < 0.0001). Conclusions: The present study demonstrated that SCZ patients had decreased expression of Kininogen-1 and genetic variants in Kininogen-1 coding gene CST9 were significantly associated with SCZ. The findings from both protein and genetic association studies suggest that Kininogen-1 could be a biomarker of SCZ.