Aimed at the energy efficiency problem with discrete power in actual wireless sensor networks, minimum energy routing algorithm is studied. Based on the typical energy consuming model, the ideal minimum energy routing is presented. Ideal minimum energy routing is a linear programming problem, and a minimum energy routing algorithm which is close to the ideal minimum energy routing algorithm--proximate optimal-point routing (POPR), is proposed. POPR is based on geographical location information and with the character of distributed and easy to be implemented in real life. The simulation results show that the performance of POPR becomes better when the node density is high.