Correlations between the air permeability coefficient and various pore structure indicators in cementitious materials were examined to determine the pore structure indicator that best evaluated air permeability using data from previous studies of air permeabilities and pore structures. The determination coefficients of air permeability with total pore volume, critical pore diameter, and ordinary threshold pore diameter were low, although these have often been used as indicators. The median and threshold pore diameters obtained by percolation theory showed high determination coefficients. The equation using the threshold pore diameter better estimated the air permeability coefficient than the Katz-Thompson equation. Keywords: air permeability, total pore volume, critical pore diameter, threshold pore diameter, median pore diameter, Katz-Thompson equation Highlights• Air permeability and pore structure of cementitious materials are closely related • Data from reports on air permeability and porosity in cements were analysed • Percolation theory-derived threshold pore diameter gives the best estimates • Proposed equation yields better results than Katz-Thompson relationship 1. Introduction The air permeability coefficient of concrete is important in evaluating the durability of concrete structures because the permeation of carbon dioxide and oxygen into concrete causes deterioration by carbonation and corrosion of reinforcement bars. Researchers have studied the relationship between the air permeability coefficient and the pore structures of concrete to understand the transportation mechanism of air and establish models that predict the air permeability of concrete; various characteristic indicators of the pore structure have been proposed. Among indicators characterising pore structure, the relationship between the total pore volume and air permeability has been reported the most extensively. The total pore volume is generally measured in one of three ways: calculation using the difference between the water-saturated and oven-dried weights [1-3], the maximum volume of cumulative intruded mercury in mercury intrusion porosimetry (MIP) [4][5][6], and image analysis of the area fraction of pores [7,8]. Some researchers set maximum and minimum pore sizes in calculating the total pore volume [9,10] without clear theoretical reason. However, as Diamond [11] noted, the pore size distribution obtained by MIP is not a true pore size distribution because of ink-bottle effects; the validity of using a total porosity with certain maximum and minimum pore sizes is unclear. Hamami et al. [12] used the main pore diameter D c , defined as the mean value of the normal distribution fitted on the main peak of a pore size distribution, as an indicator of the pore structure and reported that the square of D c multiplied by the total pore volume fraction φ showed good correlation with the air permeability coefficient. This method can be difficult to apply because the pore size distribution of concrete can be far from a normal distribut...