Perturbative Unification and Low Energy Precision Data

More precise unification predictions require going beyond the lowest order, including 2-loop running of the couplings and a correct treatment of threshold effects. Here we revised two different approaches to deal with light thresholds, based on different choices of the renormalization scheme, \bar{MS} and effective couplings. We show the equivalence of both approaches in making predictions when thresholds are taking properly into account.