Hydrogen peroxide is an environment‐friendly oxidizer, which is used in several chemical processes. However, safety necessitates the determination and control of the concentration of hydrogen peroxide during oxidation reactions. We propose a methodology to monitor hydrogen peroxide in disperse two‐phase reaction mixtures based on in‐line Raman spectroscopy. We compare indirect hard modeling (IHM), peak integration (PI), and partial least squares (PLS). Building predictive PLS and PI calibration models is challenging, whereas the IHM calibration is easy to develop. These methods show good accuracy for known samples (root mean square error of cross validation [RMSECV] of 0.3–0.7 wt %) compared to the classic titration method (RMSECV of 0.4 wt %). After calibration, inline monitoring during reaction is performed demonstrating that the concentration of hydrogen peroxide can be successfully monitored in a fast and reliable way by Raman spectroscopy. The IHM seems to give slightly better inline predictions. © 2017 American Institute of Chemical Engineers AIChE J, 63: 3994–4002, 2017