Fouling control in membrane processes is crucial to increase membrane performance and reduce costs for cleaning. An appropriate online monitoring of fouling development would allow for real time quantification and thereby for knowledge-based control strategies. However, most techniques used so far are difficult to use at industrial scale and partly limited to laboratory applications. In this study a thermal based biofilm sensor was integrated in a reverse osmosis (RO) setup treating a high organic load feed, to assess its potential as an early warning tool for permeability decline monitoring. Both membrane and sensor surface were imaged with optical coherence tomography (OCT) in parallel. The sensor surface was modified by increasing its roughness to promote biofilm attachment. The modification was effective and the sensor responded to biological growth with an increase of the signal, corresponding to a permeability decline of the RO membranes from to 1.6 to 1 L m−2 h−1 bar−1. After the system was cleaned with NaOH, 40 % of biofilm was removed and the sensor signal decreased proportionally to the restoration of permeability. The signal delivered by the sensor can therefore be used as a descriptor for biofouling severity.