A gas-liquid two-phase flow in a large diameter pipe exhibits a three-dimensional flow structure. The wire-mesh sensor (WMS) can acquire a quasi-three-dimensional void fraction distribution. Furthermore, the WMS can acquire a phasic-velocity distribution on the basis of the time lag of void signals between both sets of WMS. Previously, the acquired phasic velocity was one-dimensional distributions. The authors propose a method to estimate the three-dimensional phasic-velocity distribution from the same WMS data. A three dimensional velocity vector was determined on the basis of cross-correlation analysis. The flow direction is determined by the WMS measuring-point combination, whereby the cross-correlation coefficient between both sets of WMS measuring points reveals the peak. In addition, the flow structure can be extracted by size on the basis of a wavelet analysis. The proposed method was applied for two sets of 64 x 64 mesh sensors in an air-water flow in a vertical pipe with inner diameter of 224 mm. The proposed method can successfully visualize a swirl flow structure where large and small bubbles tend to move respectively in inward and outward directions in turn. (C) 2012 Elsevier Ltd. All rights reserved.