The study was performed over the period from June 1989 to December 1992 on 3 experimental concrete beds of 20 m2 each, constructed close to an extended aeration plant. Beds 1 and 2 were planted with reeds, bed 3 was considered as an unplanted control bed. During the first phase (June 89-May 90), most of the reeds in bed 2 died in spite of the influent sludge dose of 70 g of SS m−2 d−1. During the second phase (917 days) the 3 beds were naturally aerated from the bottom and fed with sludge that was directly extracted from the oxidation ditch (Dry Matter content 0.3%). The influent sludge load varied between 120 and 160 g of SS m−2 d−1 after the second growth season (with peak values of up to 215 g of SS m−2 d−1 in the summer of 1991). Even if bed 3 didn’t clog, the major contribution of the reeds to maintaining a high and regular liquid conductivity in the sludge has been proven, allowing easier and higher dosing of the planted beds. Monitoring of the percolation flow emphasised a high mineralization process in the rhizosphere. The accumulated sludge (Total Solids content ≈ 11%) can be dug out together with the reeds using a mechanical digger and spread on fields with a manure spreader. A regrowth of reeds occurs directly from rhizomes remaining in the drainage layer and in a few centimetres of sludge at the bottom of the bed.