Normalizing flow models have risen as a popular solution to the problem of\ndensity estimation, enabling high-quality synthetic data generation as well as\nexact probability density evaluation. However, in contexts where individuals\nare directly associated with the training data, releasing such a model raises\nprivacy concerns. In this work, we propose the use of normalizing flow models\nthat provide explicit differential privacy guarantees as a novel approach to\nthe problem of privacy-preserving density estimation. We evaluate the efficacy\nof our approach empirically using benchmark datasets, and we demonstrate that\nour method substantially outperforms previous state-of-the-art approaches. We\nadditionally show how our algorithm can be applied to the task of\ndifferentially private anomaly detection.\n