Abstract :
We compute the two-loop corrections to the coefficient of the b ← sγ magnetic penguin present in the limits of heavy top and/or heavy Higgs. These kinds of corrections affect in a significant way the observables measured at LEP. On the contrary we find that, due to a numerical accident, the correction to BR(B → Xsγ) is negligible (below the 1% level for any possible value of the Higgs mass) when the leading order result is expressed in the usual way in terms of the semileptonic BR(B → Xceν).