In this study, a nanoscale surface plasmon resonance (SPR) sensor was developed to determine the ovarian cancer marker carbohydrate antigen (CA) 125 level in serum utilizing the molecular imprinting method. For this, CA 125-imprinted poly(2-hydroxyethyl methacrylate-N-methacryloyl-(l)-tryptophan methyl ester) (p(HEMA-MATrp)) nanoparticles (NPs) were synthesized and used to coat the SPR gold chip surface. Additionally, non-imprinted p(HEMA-MATrp) nanoparticles were prepared as a control. SEM imaging, FTIR-ATR and Zeta Sizer analysis were performed for the characterization of the synthesized NPs, and the nanoscale chip surface was also characterized using BET, ellipsometry and contact angle measurements. The capability of the developed nanoscale sensor system was evaluated using the CA 125 biomarker in the concentration range of 0.1-10 U mL(-1). Ultrasensitive CA 125 glycoprotein detection was achieved with a low limit of detection value of 0.01 U mL(-1). The Langmuir adsorption model was found to be the best-suited model for expounding the interaction between the CA 125-imprinted p(HEMA-MATrp) NP chip and the CA 125 glycoprotein. In order to study the selectivity of the sensor, CA 125, human serum albumin and lysozyme were applied to the system. The highest response was obtained for CA 125. Additionally, real-sample studies were performed using artificial serum. The results showed that the developed sensor system was very effective in determining CA 125 levels even in complex media. The nanoscale chip was reused five times to detect CA 125 without any significant reduction in binding capacity, and the chip was stable for over five months. These results indicated that a precise detection method with high sensitivity, selectivity, and reliability in order to determine the level of CA 125 ovarian cancer biomarker in serum was successfully developed.