In this study, water quality and fish performance of traditional RAS with nitrifying biofilters were compared with systems operated under heterotrophic N assimilation (HET-N), a process where bacteria consume ammonium directly for growth and thereby remove dissolved N excreted from the fish, using three different modes of carbon addition. Using twelve identical pilot scale RAS, four treatment groups were established in triplicate: RAS with autotrophic biofilters (Control), RAS with autotrophic biofilters and acetate addition (BF RAS +Acetate), RAS without biofilter with acetate addition (BF RAS -Acetate) and RAS without biofilter and a biopellet reactor (Biopellet RAS). The nine RAS with carbon addition all had lower levels of nitrate and orthophosphate at the end of the trial compared to the three control RAS (approx. 70% less NO3- in the Biopellet RAS and 72% less PO43- in BF RAS -Acetate). Without biofilters installed, both BF RAS -Acetate and Biopellet RAS maintained acceptable water quality parameters during their respective start-up phases and were fully developed in under 3 weeks. The addition of acetate to the water caused an expected formation of bioflocs in the systems, and a significant increase in bacterial activity and turbidity. Substantial feed spill was observed in RAS with acetate addition. The absence of bacterial accumulation and no increase in turbidity in the water in Biopellet RAS suggest that the processes primarily occurred within the reactor. The overall fish mortality was <1%, however, both types of RAS with acetate addition led to reduced fish growth (7.4–20%) compared to the control RAS and the RAS with biopellets. Biopellets were found to reduce dissolved N and P, and had a fast start up time without deteriorating water quality, thereby showing promising traits for use in RAS.