The objective of this study was to present a prediction method for condensation heat transfer in the presence of noncondensable gas (air or nitrogen) for computational fluid dynamics (CFD) analyses, where physical quantities in the computational cells in contact with the structural wall are generally used. First, using existing temperature distributions T(y) in the turbulent boundary layer along a flat plate as functions of the distance y from the condensation surface, we evaluated the distribution of condensation heat flux qc,pre(y) from the gradient of steam concentration, we derived a modification factor η(y+) as a function of the dimensionless distance y+ to obtain a good agreement with qc,cal calculated by the qc correlation defined by using the bulk quantities; and we obtained qc,mod(y)/qc,cal = 0.90–1.10 for the region of y+ > 17. Second, we modified the local Sherwood number Sh(x) for flat plates for the boundary layer thickness δ and obtained the function Sh(δ). An existing qc correlation for flat plates as a function of Sh(δ) was applied to predict the distribution of the local value qc,pre(y), and qc,pre(y)/qc,cal = 0.95–1.15 in the best case was obtained for the region of y+ > 30. Finally, a correlation of the local Sherwood number Sh(y) was derived from the temperature distributions T(y) as a function of the local Reynolds number Re(y).