In this paper, a novel self-adaptive anode feeding pattern is introduced to optimize the output and the dynamic response of the mu DMFC (micro direct methanol fuel cell). The RSM (response surface methodology) is applied to optimize the anode flow rate, with two input parameters considered in this study, i.e., the methanol concentration and the operating current. In addition, the hardware circuit of anode self-adaptive supply system is realized based on the calculation results from RSM. Furthermore, the tests of system functionalities, including the pulse control test of micro peristaltic pump, the current collection test, and the flow rate monitoring of the self-adaptive feeding module are conducted. Finally, a metal-based mu DMFC is designed and fabricated to evaluate the output performance when the cell patterned with or without the self-adaptive system. The experimental results reveal that the novel feeding pattern can increase the output voltage under different operating conditions, and reduce the voltage response time during the current transition. (C) 2015 Elsevier Ltd. All rights reserved.