Although it has been shown that protein G based site-directed antibody capture is an advantageous immobilization technique in immunosensor development, the application of such immunosensors for direct repeated real sample measurements has not been addressed. In this surface plasmon resonance analysis based study random covalent antibody immobilization was employed as a reference technique in contrast to site-directed protein G mediated affinity capture and protein G mediated antibody capture stabilized by chemical cross-linking aimed at direct repeated human growth hormone detection in real samples. Site-directed affinity capture increased the analytical signal 2.6 times compared to random antibody immobilization. However, this approach resulted in significant antibody dissociation from the sensor surface during the measurements, which was detrimental for signal evaluation, immunosensor reusability and formed a potential platform for non-specific binding of molecules present in real samples. In contrast, chemical cross-linking of the protein G and antibody complexes under optimal conditions enabled repeated measurements and improved signal evaluation. Moreover, the analytical signal was increased 3.5 times in comparison to random antibody immobilization. The obtained results put emphasis on sensing molecule orientation, stability surface mass density and remaining activity after the immobilization. Based on these findings an immunosensor model for human growth hormone detection was developed. The limit of direct detection of the developed immunosensor was 0.99 nM, the limit of quantification was 3.31 nM and the linear detection range was from 3 to 9 nM. It has been demonstrated that the immunosensor model was prone to non-specific binding of serum proteins that could be minimized following a one-step sample pre-treatment procedure.