The importance of probability-based approaches for duplicate detection has been recognized in both research and practice. However, existing approaches do not aim to consider the underlying real-world events resulting in duplicates (e.g., that a relocation may lead to the storage of two records for the same customer, once before and after the relocation). Duplicates resulting from real-world events exhibit specific characteristics. For instance, duplicates resulting from relocations tend to have significantly different attribute values for all address-related attributes. Hence, existing approaches focusing on high similarity with respect to attribute values are hardly able to identify possible duplicates resulting from such real-world events. To address this issue, we propose an approach for event-driven duplicate detection based on probability theory. Our approach assigns the probability of being a duplicate resulting from real-world events to each analysed pair of records while avoiding limiting assumptions (of existing approaches). We demonstrate the practical applicability and effectiveness of our approach in a real-world setting by analysing customer master data of a German insurer. The evaluation shows that the results provided by the approach are reliable and useful for decision support and can outperform well-known state-of-the-art approaches for duplicate detection.