In the arid climate of Djibouti, salt concentrations of up to 1200 mg L −1 have been recorded in well water, attributed to the overexploitation of aquifers. Farmers are therefore turning to untreated wastewater as a source of water for crop irrigation. The authors undertook a study in two phases. In the first phase they tested the field transposition of an experimental water treatment method previously laboratory-tested on a small scale. The pilot device was composed of an adaptor basin for the wastewater, a filter basin filled with sand and a storage basin to collect and hold the treated water. Transfer of waters from basin to basin was gravitational. The physicochemical and microbiological sleek raw water parameters were determined weekly for 7 weeks. Monitoring of chemical oxygen demand concentrations decreased from 445 ± 5.76 mg L −1 O 2 on the first day of the experiment to 101 ± 1.34 mg L −1 O 2 on the 21st day. The removal of total coliforms attained more than 99.6% as of day 21. The performance of the microbiological treatment of waters obtained through sand filtration is therefore of definite interest in agriculture. In the second phase the effect of different irrigation waters (untreated wastewater (WW), treated wastewater (TW), saline well water (SW), mixed water (MW) and potable water (PW)) on growth and yield of the grass Panicum maximum were also compared. The use of WW and TW in irrigation effectively increased stem height and aerial part dry matter compared to other treatments, while irrigation with SW and PW showed the lowest measured values. Depending on their performance, the effectiveness of the treated waters can be ranked in the following increasing order: TW; WW > MW > PW; SW. The performance of the TW treatment can be explained by its higher nutritive content, especially in N.