The thesis shows how second-best mechanisms (labour markets rigidities, imperfect foresight) and mechanisms considered as short-term issues (inertia of productive capital,unemployment) induce that (i) mitigation costs are transition costs insofar as policies to reorient investments to low-carbon infrastructures are implemented early to avoid lock-ins in high-carbon development paths ; (ii) these transition costs can be significant, in particular for emerging and developing countries. Therefore, representing real world rigidities in energy-economy-environment models has important implications for the analysis of climate policies. Equalizing carbon prices is not anymore the policy minizing mitigation costs if a second-best world is considered...