Micropollutants are regularly detected at the outlets of wastewater treatment plants (WWTPs). Across urban and industrial WWTPs, monitoring directives only require assessment for a handful of chemicals via sampling methods that fail to capture the temporal variability in micropollutant discharge. In this study, we develop a biotest for real-time on-line monitoring of micropollutant discharge dynamics in WWTPs effluents. The selected biomonitoring device ToxMate uses videotracking of invertebrate movement, which was used to deduce avoidance behaviour of the amphipod Gammarus fossarum. Organism conditioning was set up to induce a state of minimal locomotor activity in basal conditions to maximise avoidance signal sensitivity to micropollutant spikes. We showed that with a standardised protocol, it was possible to minimise both overall movement and sensitivity to physio-chemical variations typical to WWTP effluents, as well as capture the spikes of two micropollutants upon exposure (copper and methomyl). Spikes in avoidance behaviour were consistently seen for the two chemicals, as well as a strong correlation between avoidance intensity and spiked concentration. A two-year effluent monitoring case study also illustrates how this biomonitoring method is suitable for real-time on-site monitoring, and shows a promising non-targeted approach for characterising complex micropollutant discharge variability at WWTP effluents, which today remains poorly understood.