Osmotic backwash experiments were conducted and analytical model was developed in attempt to describe the backwash mechanism. A single unit of a spiral wound RO membrane was used in the experiments. The RO membrane was fed by salted water solution. Permeate flux characteristics of the membrane were determined through steady-state RO experiments. Then the system was shifted immediately to a backwash process by reducing the feed pressure, Delta p, either to zero or to a level below the osmotic pressure to allow net backwash driving force. The backwash experiments reveal that the backwash process has two distinct regions. The flow rate drops sharply at the initial backwash process, followed by a prominently slower flow rate continuously slows down until it reaches a constant value (toward zero, for Delta p = 0). These results suggest that the first backwash stage acts mainly to dilute the salt concentration at the feed concentration polarization (CP) layer. The second stage of the backwash flow rate exhibits salt dilution of the bulk solution. RO experiments were conducted also with a super-saturated CaCO3 solution, to cause salt precipitation and partially clogging of the membrane surface followed by flux reduction. The permeate flux was resumed to its original level with osmotic backwash cleaning of the membrane. Effects of three independent RO feed variables; feed concentration, flow rate, and applied pressure, on the accumulated volume of backwash water, v(t) were analyzed experimentally. It was found that feed concentration has the strongest effect on v(t), while the other two parameters has only minor effects on the process. Presence of operational pressure during the backwash process reduces v(t) dramatically, as a consequence of the driving force reduction. A simple analytical model was developed and fits well the experimental data of the second stage without feed flow during the backwash process.