Research has shown that NPD project leaders should engage in boundary-spanning activities. The present study tested the impact of four boundary-spanning activities on NPD project performance and analyzed the antecedents of these activities. We hypothesized that NPD project leaders' abilities to perform these activities depend on the characteristics of their personal networks -- structural holes, strength of ties, vertical and horizontal bridging ties. A Partial Least Squares test on 73 NPD projects showed that (a) "obtaining political support" and "scanning for ideas" are the boundary activities with the greatest impact on performance, (b) project leaders with strong ties in their network are more effective at these activities, (c) project leaders with structural holes in their networks are more effective in another boundary activity, "protecting the team", although this activity does not affect NPD outcomes. These results represent an important contribution to understanding how team leaders contribute to project performance.