Abstract: In this paper, a directivity corrected phase coherent factor (DC_PCF) weighted total focusing imaging algorithm is proposed to address the ultrasonic attenuation problem in thick-walled welds of 9%Ni steel, which caused by coarse grains and inhomogeneous anisotropy of crystals. Experimental results of testing defects in a 30.0 mm thick 9Ni steel T-shaped fillet welds specimen (inner hole of weld, hole at face groove and groove at face groove), demonstrate that structural noise is reduced and defect recognition is improved. The average error of depth detection for defects is 2.6 mm, and the average error of width detection for defects is 0.2 mm, which are both smaller than those obtained with the total focusing method (TFM) and phase coherence imaging (PCI) algorithm. Compared to the TFM, detection signal to noise ratios of three defects with the proposed algorithm are improved by 2.1 dB, 5.1 dB and 4.9 dB respectively; compared to the PCI algorithm, the improvements are 4.6 dB, 3.5 dB and 4.1 dB respectively.