Constructed wetlands are considered an inefficient technology for long-term phosphorus (P) removal. The P retention effectiveness of subsurface wetlands can be improved by using appropriate substrates. The objectives of this study were to: (i) use sorption isotherms to estimate the P sorption capacity of the two materials, masonry sand and light weight expanded shale; (ii) describe dissolved P removal in small (2.7 m3) subsurface flow wetlands; (iii) quantify the forms of P retained by the substrates in the pilot cells; and (iv) use resulting data to assess the technical and economic feasibility of the most promising system to remove P. The P sorption capacity of masonry sand and expanded shale, as determined with Langmuir isotherms, was 60 mg/kg and 971 mg/kg respectively. In the pilot cells receiving secondarily treated wastewater, cells containing expanded shale retained a greater proportion of the incoming P (50.8 percent) than cells containing masonry sand (14.5 percent). After a year of operation, samples were analyzed for total P (TP) and total inorganic P (TIP). Subsamples were fractionated into labile-P, Fe+Al-bound P, humic-P, Ca+Mg-bound P, and residual-P. Means and standard deviations of TP retained by the expanded shale and masonry sand were 349 + 169 and 11.9 …