In this study, two methods were examined to optimize the immobilization of antibodies on paper when conducting a paper-based enzyme-linked immunosorbent assay (P-ELISA). Human IgG, as a test-capture protein, was immobilized on paper via the formation of Schiff bases. Aldehyde groups were introduced onto the surface of the paper via two methods: NaIO4 and 3-aminopropyltriethoxysilane (APTS) with glutaraldehyde (APTS-glutaraldehyde). In the assay, horseradish peroxidase-conjugated anti-human IgG (HRP-anti-IgG) binds to the immobilized human IgG, and the colorimetric reaction of 3,3',5,5'-tetramethylbenzyzine (TMB) produces a blue color in the presence of H2O2 and HRP-anti-IgG as a model analyte. The immobilization of human IgG, the enzymatic reaction conditions, and the reduction of the chemical bond between the paper surface and immobilized human IgG all were optimized in order to improve both the analytical performance and the stability. In addition, the thickness of the paper was examined to stabilize the analytical signal. Consequently, the APTS-glutaraldehyde method was superior to the NaIO4 method in terms of sensitivity and reproducibility. Conversely, the reduction of imine to amine with NaBH4 proved to exert only minimal influence on sensitivity and stability, although it tended to degrade reproducibility. We also found that thick paper was preferential when using P-ELISA because a rigid paper substrate prevents distortion of the paper surface that is often caused by repeated washing processes.