Experimental bounds on induced vacuum magnetic birefringence can be used to improve present photon-photon scattering limits in the electronvolt energy range. Measurements with the PVLAS apparatus (E. Zavattini {it et al.}, Phys. Rev. D {bf77} (2008)
032006) at both $lambda = 1064$ nm and 532 nm lead to bounds on the parameter {it A$_{e}$}, describing non linear effects in QED, of $A_{e}^{(1064)} < 6.6cdot10^{-21}$ T$^{-2}$ @ 1064 nm and $A_{e}^{(532)} < 6.3cdot10^{-21}$ T$^{-2}$ @ 532 nm, respectively, at 95% confidence level, compared to the predicted value of $A_{e}=1.32cdot10^{-24}$ T$^{-2}$. The total photon-photon scattering cross section may also be expressed in terms of $A_e$, setting bounds for unpolarized light of $sigma_{gammagamma}^{(1064)} < 4.6cdot10^{-62}$ m$^{2}$ and $sigma_{gammagamma}^{(532)} < 2.7cdot10^{-60}$ m$^{2}$. Compared to the expected QED scattering cross section these results are a factor of $simeq2cdot10^{7}$ higher and represent an improvement of a factor about 500 on previous bounds based on ellipticity measurements and of a factor of about $10^{10}$ on bounds based on direct stimulated scattering measurements.
