In this paper, a three-dimensional solid-fluid conjugate model coupled with a simplified conjugate-gradient method was employed to optimize the performance of double-layered microchannel heat sinks. Channel number, channel width, bottom channel height, and bottom coolant inlet velocity were selected as search variables to achieve the optimal heat sink performance. Firstly, two single-objective optimizations based on different objective functions (one is the maximum temperature change on the bottom wall Delta T-w,T-b and the other is the overall thermal resistance R) were performed at a constant pumping power. Subsequently, the effects of total pumping power on the optimal Delta T-w,T-b and R were analyzed, and the optimal search variables at various pumping powers were obtained. For single-objective optimization with the objective function of Delta T-w,T-b, Delta T-w,T-b is respectively decreased by 6.01, 5.29, and 2.99 K when compared with three original designs. For the objective function of R, however, R is respectively decreased by 36.51%, 15.10%, and 16.67%. The results also indicate that R and Delta T-w,T-b cannot achieve their optimal values simultaneously by the two single-objective optimizations. Thus, a multi-objective optimization was carried out, which demonstrates that when a set of desirable values of Delta T-w,T-b and R is required by designers,. the present multi-objective optimization could meet this requirement. (C)2015 Elsevier Ltd. All rights reserved.