Critics of the proposed Transatlantic Trade and Investment Partnership (TTIP) dismiss its potential welfare gains as small compared with its risks. We contribute to this debate by investigating the driving forces behind the magnitudes of the estimated welfare gains using the structurally estimated general equilibrium trade model by Egger and Larch (2011) for 173 countries. In our baseline scenario, the TTIP amounts to a reduction of ad valorem trade costs across the Atlantic between 16 and 26 percentage points. We find that the TTIP could yield substantial gains for the EU (3.9%), the United States (4.9%), and the world (+1.6%). While welfare gains are heterogeneous within the EU, the TTIP does not systematically favour richer or more central member states. The majority of third countries would be negatively affected (0.9% on average). We identify as key drivers for the magnitudes of the welfare effects different assumptions about trade cost specifications, about the assumed trade cost reducing potential of the TTIP, about different levels of aggregation, and about the regulatory spill-overs of the TTIP on third countries. Our insights on the drivers for the welfare effects help to understand differences across current evaluations of the TTIP.