A laboratory study is reported regarding fatigue damage growth monitoring in a complete 45.7 m long wind turbine blade typically designed for a 2 MW generator. The main purpose of this study was to investigate the feasibility of in-service monitoring of the structural health of blades by acoustic emission (AE). Cyclic loading by compact resonant masses was performed to accurately simulate in-service load conditions and 187 kcs of fatigue were performed over periods which totalled 21 days, during which AE monitoring was performed with a 4 sensor array. Before the final 8 days of fatigue testing a simulated rectangular defect of dimensions 1 m x 0.05 m x 0.01 m was introduced into the blade material. The growth of fatigue damage from this source defect was successfully detected from AE monitoring. The AE signals were correlated with the growth of delamination up to 0.3 m in length and channel cracking in the final two days of fatigue testing. A high detection threshold of 40 dB was employed to suppress AE noise generated by the fatigue loading, which was a realistic simulation of the noise that would be generated in service from wind impact and acoustic coupling to the tower and nacelle. In order to decrease the probability of false alarm, a threshold of 45 dB was selected for further data processing. The crack propagation related AE signals discovered by counting only received pulse signals (bursts) from 4 sensors whose arrival times lay within the maximum variation of travel times from the damage source to the different sensors in the array. Analysis of the relative arrival times at the sensors by triangulation method successfully determined the location of damage growth, which was confirmed by photographic evidence. In view of the small scale of the damage growth relative to the blade size that was successfully detected, the developed AE monitoring methodology shows excellent promise as an in-service blade integrity monitoring technique capable of providing early warnings of developing damage before it becomes too expensive to repair. (C) 2016 Elsevier Ltd. All rights reserved.