Drawing on insights from network dynamics and exchange theory, I develop and test arguments for the retention or dissolution of exchange relationships. I exploit mergers and acquisitions among advertising firms as strategic actions that change the networks in which they and their clients are situated, and examine the consequences of these changes for their network relationships. Analysis of an archival, longitudinal data set confirms that, in general, relational embeddedness reduces the likelihood of dissolution and that increases in competitive overlap among clients increase dissolution likelihood. The results also provide evidence of a significant interaction effect between relational embeddedness and competitive overlap. For low to moderate increases in competitive overlap, embeddedness reduces dissolution likelihood. However, when the merger results in a high increase in competitive overlap, increasing embeddedness actually increases dissolution likelihood. Mechanisms to explain the findings are explored, including fears of information leakage and trust betrayal. The findings suggest that under certain conditions, relational embeddedness can reduce-rather than increase-relationship stability.