Two-loop contribution to the pion transition form factor vs. experimental data

We present predictions for the pion-photon transition form factor, derived with the help of light-cone sum rules and including the main part of the NNLO radiative corrections. We show that, when the Bakulev-Mikhailov-Stefanis (BMS) pion distribution amplitude is used, the obtained predictions agree well with the CELLO and the CLEO data. We found that no model distribution amplitude can reproduce the observed $Q^2$ growth of the new BaBar data, though the BMS model complies with several BaBar data points.