This paper considers a heterogeneous network (HetNet), which consists of one macro base station (MBS) and numerous small cell base stations (SBSs) cooperatively serving multiple user terminals. The first objective is to design cooperative transmit beamformers at the base stations to maximize the network energy efficiency (EE) in terms of bits per Joule subject to the users’ quality of service (QoS) constraints, which poses a computationally difficult optimization problem. The commonly used Dinkelbach-type algorithms for optimizing a ratio of concave and convex functions are not applicable. The paper develops a path-following algorithm to address the computational solution to this problem, which invokes only a simple convex quadratic program of moderate dimension at each iteration and quickly converges at least to a locally optimal solution. Furthermore, the problem of joint beamformer design and SBS service assignment in the threeobjective (EE, QoS and service loading) optimization is also addressed. Numerical results demonstrate the performance advantage of the proposed solutions.