The paper deals with a treatment for obtaining phosphorus recovery from anaerobic supernatants with low phosphate concentration (30-50 mg PO4-P/L). The study was carried out at demonstrative scale (inflow up to 2 m(3)/d), and operative conditions as fluidized or packed bed were investigated. To determine the best operative conditions in the fluidized bed reactor, two upward velocities of 42 and of 64 m/h were performed. Very high phosphorus removal (up to 75%) was observed, and the real performances were similar to those predicted by the mathematical model. Regarding the fluidized bed operative condition into the reactor, the average value of V'10 was maintained at 110 mL/L throughout the treatment of 234 m,9 of anaerobic supernatants. In the second part of the experiment, the polyelectrolyte escape from the dewatering station caused the packing of a great part of the bed, and only a minimum content of the grains remained fluidized (20 000 mg/L). Good performances in phosphorus removal were observed (45-55%), but they were lower than those expected in the fluidized bed (64-69%). Big crystal dimensions in the bottom of the reactor were shown in the packed bed (phi(50) = 0.5-1.4 mm), whereas they maintained a dimension up to 0.5 mm in the fluidized bed. Suspended grains showed a dimension strictly related to the recycle flow rate (phi(50) = 0.21 mm at 12 m(3)/h and -phi(50) = 0.35 mm at 18 m(3)/h). The phosphorus mass balance showed a production of similar to 60 kg of grains in the fluidized bed and 111 kg in the packed bed during the experimental periods. Both were mainly constituted by struvite (97%) with smaller amounts of calcite and hydroxyapatite. A low content of organic substances in phosphorus salts extracted from the bottom of the reactor was quantified by a COD to a TS ratio of 2-3%.