This field-scale hydroponic experiment investigated the effects of plant density and nutrient levels on arsenic (As) removal by the As-hyperaccumulator Pteris vittata L. (Chinese brake fern). All ferns were grown in plastic tanks containing 30 L of As-contaminated groundwater (130 microg x L(-1) As) collected from South Florida. The treatments consisted of four plant densities (zero, one, two, or four plants per 30 L), two nitrogen (N) concentrations (50% or 100% of 0.25-strength Hoagland solution [HS]), and two phosphorous (P) concentrations (15% and 30% of 0.25 strength HS). While low P was more effective than high P for plant As removal initially, N levels showed little effect. At 15% P, it took 3 wk for the ferns at a plant density of four to reduce As to less than 10 microg L(-1) (USEPA and WHO standard), whereas it took 4-6 wk at plant densities of one or two. For reused ferns, established plants with more extensive roots than "first-time" ferns, a low plant density of one plant/30 L was more effective, reducing As in water to less than 10 microg L(-1) in 8 h. This translates to an As removal rate of 400 microg h(-l) plant(-1), which is the highest rate reported to date. Arsenic-concentration in tanks with no plants as a control remained high throughout the experiment. Using more established ferns supplemented with dilute nutrients (0.25 HS with 25% N and 15% P) with optimized plant density (one plant per 30 L) reduced interplant competition and secondary contamination from nutrients, and can be recommended for phytofiltration of As-contaminated groundwater. This study demonstrated that P. vittata is effective in remediating As-contaminated groundwater to meet recommended standards.