Group estimation of user stories is an important part of Extreme Programming (XP), used for both planning releases and iterations. Research has shown that although group estimation in many cases is superior to individual estimation, there is still room for improvement. For instance, group estimation performance can be reduced by dominant personalities and anchoring effects. Through the analysis of 101 user story estimates, made by an XP team for release planning, we investigate whether the introduction of the planning poker estimation process improved the estimation ability of the team. The results show that planning poker improved the team's estimation performance in most cases, but that it increased estimation error in the extreme cases