A hybrid anchor-free localisation scheme for multihop wireless sensor networks is presented. First, a relatively dense group of nodes is selected as a base, which are localised by using the multidimensional scaling method. Secondly, the robust quads (RQ) method is employed to localise other nodes, following which the robust triangle and radio range (RTRR) approach is used to perform the localisation task. The RQ and the RTRR methods are used alternately until no more nodes can be localised by the two approaches. Simulation results demonstrate that the proposed hybrid localisation algorithm performs well in terms of both accuracy and the success rate of localisation. To evaluate the accuracy of anchor-free localisation algorithms, the authors derive two different accuracy measures: the Cramer-Rao lower bound (CRLB) to benchmark the coordinate estimation errors and the approximate lower bound to benchmark the distance errors. Simulation results demonstrate that both the CRLB and the distance error lower bound provide references for the accuracy of the location algorithms.