In the experiment with the SND detector at the VEPP-2M e + e − collider measuring the e + e − → ηγ cross section in the energy range 0.6 < √ s < 1.38 GeV the reanalysis of data is performed. The goal is to improve the accuracy of the previous results by analysing ambiguities in the approximation of energy dependence of the e + e − → ηγ cross section, which were not taken into account in our previous study. We report new results on the approximation of the e + e − → ηγ cross section based on Vector Dominance Model under new model parameter assumptions. In our previous work [1] we performed a measurement of the e + e − → ηγ reaction cross section for two η-meson decay modes: η → 3π 0 and η → π + π − π 0 . The measured cross section was approximated within Vector Dominance Model (VDM). In this approximation the fact that with the free magnitudes and phases of all four ρ, ω, φ, and ρ ′ resonances, the minimized likelihood function has several minima and thus, a choice of the right minimization solution is ambiguous, was ignored. Additional physical restrictions must be applied to, at least, the parameters of the ρ ′ resonance, in order to make a right choice of the minimization solution.In this work the ρ ′ -meson parameters were estimated from the experimental data on the e + e − → ηρ cross section [2,3]. The cross section value in the ρ ′ -resonance maximum obtained from the approximation of the e + e − → ηρ cross section can be translated to the e + e − → ρ ′ → ηγ cross section using VDM as in Ref. [4]. The approximation of the data on the e + e − → ηρ cross section was done with three different models describing the energy dependence of the ρ ′ resonance width:Under these assumptions the resulting mass, width, and maximum cross section of the ρ ′ resonance varied within 1440-1520 MeV, 220-400 MeV, and 0.08-0.13 nb, respectively. All three ρ ′ parameter sets were used in the e + e − → ηγ cross section approximation. The difference in the experimental results was included in the systematic uncertainty of the final result. The cross section at the ρ ′ resonance maximum acquired an additional 20% systematic error from the uncertainty of the cross section calculation in the VDM.Natural values for the phase of the ρ ′ amplitude are 0 or 180 degrees. The phase shift from these values can be induced by the ρ and ρ ′ mixing via common decay channels, for example, ρ, ρ ′ → ωπ 0 . In the model considered in Ref. [5] it was shown that mixing is strongly energy dependent. For the approximation of the data on e + e − → ηγ cross section the ϕ ρ ′ value at the φ resonance maximum is important. Since at this energy the ρ − ρ ′ mixing is small, no significant phase shifts from 0 or 180 degrees are expected. Two approximations with the ϕ ρ ′ phase fixed at these values were considered. In order to estimate a corresponding systematic uncertainty of the approximation parameters due to the ρ ′ phase uncertainty, we varied ϕ ρ ′ by ±20 • with respect to its initial values.The fitting results with the ρ ′ parameters fixed to described ...