The Green Roof Water recycling system (GROW) was established at Teachers hostel, Mehran UET. Jamshoro. Grey water generated at hostel was treated through above said system. The operating parameters i.e flow rate and HRT were optimized. The various local plant species were planted in the GROW system and their contaminant uptake/ removal efficiency was measured by analyzing influent and effluent COD, BOD and TSS respectively. This study finds the GROW System as most suitable technology to treat greywater in hot climates. The grey water flow rate for GROW system was optimized as 0.720 m3.d-1with HRT of one day giving both better pollutant removal and green water yield. The optimized air flow rate was also found as 5.0 Liters/min., giving maximum pollutants elimination efficacies for BOD5, COD and TSS as 77%,73% and 58% respectively. Influent greywater was characterized in terms BOD5, 107±38 mg/L, COD 235.8±66 mg/Land TSS 96±45 mg/L These values were found close to the literature cited. GROW System performance under mixed mode plantation gave promising results in removal of BOD, COD and TSS as 78.2±11%, 64.9±10.7% and 78.89±16.8% respectively. The influent grey water was medium category COD: BOD ratio ranged 0.4-0.6, which is indicative of medium to high biodegradability The indigenous Ninthra (Mentha requienii) and Mothapalm or River lily (Crinum pedunculatum)) plant species gave highest results in removal of pollutants. The rice (Oriza/satival) specie also performed well, whereas, the Water-Hyacinth specie was very successful in removing COD upto 47%. The combined effect of all species grown separately in row 2 – 5 including first row of filter bed gave excellent performance of GROW system as BOD5, COD and TSS removal up to 82%, 93.5% and 85% respectively. The treated effluent concentration in terms of BOD5, COD and TSS are 19.9 mg/L, 71.9 mg/L and 18.40 mg/L respectively, which is far less than the Sindh Environmental Quality Standards (SEQS ,2016), for disposal of treated waste water in to inland waters. This water can be used for toilet flushing and home gardening, if handled with care. A new GROW Model was developed and validated with the experimental data, this model predicts effluent concentration of BOD, COD and TSS more accurately as compared with the Kinetic Model. Finally, the design guidelines are also proposed for the application of GROW Technology in the urban settings.