This paper addresses two questions: (i) Are DP-internal head-modifier agreement and subject-verb agreement instances of the same phenomena occurring in different syntactic domains, and (ii) How can we use existing theoretical machinery to analyze systems of concord? Based on evidence from distribution in Icelandic, I answer question (i) in the negative, and then develop a novel analysis of the concord system in Icelandic to respond to question (ii). Building on research in the framework of Distributed Morphology, I argue that features from concord are collected in one location in the narrow syntax and then distributed from that location to the heads showing concord at PF. I conclude by considering what concord looks like outside of Icelandic and what nominal agreement looks like outside of concord.