The flow structure in a developing air-water two-phase flow was investigated experimentally along a large vertical pipe (inner diameter, D-h : 0.48 m, ratio of length of flow path L to D-h : about 4.2). Two air injection methods (porous sinter injection and nozzle injection) were adopted to realize an extremely different Bow structure in the developing region. The Bow rate condition in the test section was as follows : superficial air velocity : 0.02-0.87 m/s (at atmospheric pressure) and superficial water velocity : 0.01-0.2 m/s, which covers the range of bubbly to slug Bow in a small-scale pipe (D-h less than or equal to about 0.05 m). No air slugs occupying the flow path were recognized in this experiment regardless of the air injection methods even under the condition where slug Bow is realized in the small-scale pipe. In the lower half of the test section, the axial distribution of sectional differential pressure and the radial distribution of local void fraction showed peculiar distributions depending on the air injection methods. However, in the upper half of the test section, the effects of the air injection methods are small in respect of the shapes of the differential pressure distribution and the phase distribution. The comparison of sectional void fraction near the top of the test section with Kataoka’s correlation indicated that the distribution parameter of the drift-flux model should be modeled including the effect of D-h and the bubble size distribution is affected by the air injection methods. The bubble size distribution is considered to be affected also by L/D-h based on comparison of results with Hills’ correlation.