How to achieve transmissions in an energy-efficient way in multi-hop decode and forward (DF) relay cognitive radio sensor networks (CRSNs) is important since sensor nodes in CRSNs are usually battery powered. This paper aims to maximize energy efficiency (EE) by joint optimizing sensing time and power allocation in multi-channels & multi-hops DF relay CRSNs under constraints on outage probability and sensing performance. First, we design a channel selection scheme for sensing according to the available probabilities of multi channels. Second, we analyze the expected throughput and energy consumption and formulate the EE problem as a concave/ concave fractional program. Third, coordinate ascent and Charnes-Cooper Transformation (CCT) methods are used to transform the nonlinear fractional problem into an equivalent concave problem. Subsequently, the closed form of outage probability is derived and the convergence rate of the iterative algorithm is analyzed. Finally, simulation results show that the proposed scheme can achieve effective EE.