Introduction Further development of recirculating aquaculture systems (RAS) towards zero-exchange depends mostly on the improvement of water treatment technologies. Ozone and UV radiation are leading technologies requiring high energy demand and educated staff able to manage the hurdles of their application. Moreover, there are some constrains for systems with poor mechanical filtration or where accumulation of particles higher than 50 µm significantly reduce the penetration potential of UV application. An alternative method commonly used in wastewater treatment to eliminate particulate aggregates is sonication. This method is based on cavitation effects which contribute to disrupt bacterial bioflocs and to break microbial cell walls leading to reduced viability. The present study aims to evaluate the disinfection capacity of a prototype created to treat process water in a RAS rearing aquaculture relevant freshwater and saltwater species with three sonication frequencies. The potential impact on the microbiome of the system in different compartments beside the reactor as well as bacterial viability was evaluated. Material and Methods An ultrasound prototype composed of 12 inducers connected to control devices was created in the frame of this project and adapted to a 5 m3 research RAS composed of three rearing tanks, a drum filter, 2 biofiltration units (nitrification denitrification), a sump and a protein skimmer with ozone disinfection. For the experiments the system was initially prepared for rearing European Seabass (Dicentrarchus labrax) and in a second experiment for rearing tilapia (Oreochromis niloticus). Process water coming from the sump was conducted into the prototype at a flow rate of 10 l/min and treated with 575 kHz, 862 kHz and 1142 kHz without further disinfection. For the saltwater experiments we tested 50% and 75% frequencies amplitude while only 75% amplitude was used for freshwater experiments. Each frequency was applied for 96 h and daily sampling was conducted to determine variations on microbial viability (BacLight Viability Kit) and bacterial community composition (FISH) with respect to reference samples collected before treatment. For FISH analysis (Fig. 1) generic FAM labelled DNA probes for Eubacteria (EUB) and Archaea (ARCH) as well as more specific probes for α-, β-, γ-, δ-Proteobacteria (ALF, BET, GAM, DELTA) and Actinobacteria (HGC) were included. When available, also non labelled competitor DNA probes in equimolar concentration as the respective labelled probe were used. All samples were counterstained with DAPI. Results and Discussion Marine RAS: The sterilizing effect was impacted by the amplitude used. Frequencies 575 kHz and 1142 kHz showed higher disinfection potential by 75% amplitude than 50%. The proportion of dead cells increased with the frequency. At 1142 kHz, a decrease in the total number of most of the selected bacterial groups was detected (Fig. 2) while the total numbers of bacteria at the end of application did not significantly change when using 575 kHz and 860 kHz. Sonication with all tested frequencies lead to changes in the bacterial community. Especially at 1142 kHz, a strong decrease in ALF, BET, GAM and ARC and an increase in DEL was observed (Table 1). This suggests a selective effect of US treatment on microbial community. Freshwater RAS: The sonication treatment of the system while rearing freshwater species did not show a defined impact with respect to changes in bacterial composition over time. At a frequency of 860 kHz, there was an increase in the number of counted bacteria over time (Fig. 3) while a slightly drop was observed by 575 and 1142 kHz. No marked changes in the composition of the bacterial community were detected for the latter frequencies. For all frequencies tested there was no conspicuous change in the percentage of dead cells.