Seaweed biofilters have proven their usefulness in the treatment of fishpond effluents. However, their performance poses a dilemma: TAN (Total Ammonia N) uptake rate – and with it seaweed yield and protein content – is inversely proportional to TAN uptake efficiency. The ideal for a seaweed biofilter performance would be a high uptake rate together with high uptake efficiency. The novel three-stage seaweed biofilter design described here has solved this dilemma. The design used the finding that the performance of seaweed ponds depended on the flux of TAN through them, and that therefore effluents with reduced TAN concentration could provide the seaweed with a high TAN flux if the water flow increased proportionally. Effluents from a seabream fishpond were passed through a series of three successively smaller (25, 12.5 and 6.25 m2, respectively) air-agitated Ulva lactuca ponds. The diminished inflow TAN concentrations to the second and third ponds of the biofilter system were compensated for by the increased water exchange rates, inversely proportional to their sizes. The biofilter performance was evaluated under several TAN loads. TAN was efficiently removed (85–90%), at a high areal rate (up to 2.9 g N m−2 d−1) while producing high protein U. lactuca (up to 44% dw) in all three stages, although with mediocre yields (up to 189 g fresh m−2 d−1). Performance of each seaweed biofilter pond correlated not with TAN concentration, but with areal TAN loads. The novel three-stage design provides significant functional and economic improvements in seaweed biofiltration of intensive fishpond water.