This work develops a radio resource allocation framework for energy-efficient and confidential communications in multiple-antenna wiretap channels. Statistical channel state information (CSI) is considered at the transmitter, and the artificial noise (AN) technique is used. The formulated problem is a challenging non-convex fractional program which is tackled by combining fractional programming and sequential optimization. The resulting sequential fractional programming algorithm exhibits polynomial complexity and is provably convergent to a first-order optimal resource allocation. Numerical results show the merits of the proposed algorithm with AN.