Context : \gamma Peg is a bright B pulsator showing both p and g modes of \beta Cep and SPB types .
It has also been claimed to be a magnetic star by some authors while others do not detect a magnetic field .

Aims : We aimed at checking for the presence of a magnetic field , characterise it if it exists or provide a firm upper limit of its strength if it is not detected .
If \gamma Peg is magnetic as claimed , it would make an ideal asteroseismic target to test various theoretical scenarios .
If it is very weakly magnetic , it would be the first observation of an extension of Vega-like fields to early B stars .
Finally , if it is not magnetic and we can provide a very low upper limit on its non-detected field , it would make an important result for stellar evolution models .

Methods : We acquired high resolution , high signal-to-noise spectropolarimetric Narval data at Telescope Bernard Lyot ( TBL ) .
We also gathered existing dimaPol spectropolarimetric data from the Dominion Astrophysical Observatory ( DAO ) and Musicos spectropolarimetric data from TBL .
We analysed the Narval and Musicos observations using the LSD ( Least-Squares Deconvolution ) technique to derive the longitudinal magnetic field and Zeeman signatures in lines .
The longitudinal field strength was also extracted from the H \beta line observed with the DAO .
With a Monte Carlo simulation we derived the maximum strength of the field possibly hosted by \gamma Peg .

Results : We find that no magnetic signatures are visible in the very high quality spectropolarimetric data .
The average longitudinal field measured in the Narval data is B _ { l } = -0.1 \pm 0.4 G. We derive a very strict upper limit of the dipolar field strength of B _ { pol } \sim 40 G .

Conclusions : We conclude that \gamma Peg is not magnetic : it does not host a strong stable fossil field as observed in a fraction of massive stars , nor a very weak Vega-like field .
There is therefore no evidence that Vega-like fields exist in B stars contrary to the predictions by fossil field dichotomy scenarios .
These scenarios should thus be revised .
Our results also provide strong constraints for stellar evolution models .