We study an open text mining problem – discovering concept-level event associations from a text stream. We investigate the importance and challenge of this task and propose a novel solution by using event sequential patterns. The proposed approach can discover important event associations implicitly expressed. The discovered event associations are general and useful as knowledge for applications such as event prediction.