Sensor network security has received much deserved attention in the last few years. Key management is at the heart of securing such networks. Key management schemes for sensor networks may be classified broadly into static and dynamic keying based on administrative key updates after network deployment. While static schemes assume no updates, dynamic ones provide for post-deployment administrative key updates. The general security and performance objectives of key management schemes include minimizing number of keys stored per sensor node, providing rich logical pairwise connectivity, and enhancing network resilience to node capture. The contribution of this paper is two-fold. First, we propose an efficient dynamic key management scheme that uses key polynomials and is based on Exclusion Basis Systems (EBS); a combinatorial formulation of the group key management problem. Second, we compare our scheme to both static and dynamic keying with regards to resilience to node capture, overhead, and connectivity. Up to our knowledge, this is the first study of dynamic versus static keying schemes in sensor networks. Our results show that dynamic keying using EBS-based, as opposed to random, key assignment together with efficient rekeying mechanisms may lead to better security and may outperform static keying.