A real-world validation dataset collected to evaluate the sim-to-real transfer capabilities of event-based optical flow algorithms trained on synthetic data, more specifically, the synthetic eStonefish-Scenes dataset. Acquired in a controlled underwater environment, it offers a rigorous benchmark for assessing algorithm performance under real sensing conditions. Experimental Setup & Features Platform: Data was collected using a BlueROV2 equipped with a DAVIS346 event camera. Environment: The ROV navigated over a large, high-resolution textured poster placed on the floor of the CIRS indoor testing pool in Girona, Spain. Motion: The dataset includes diverse 6-DOF motion patterns, including translation and rotation, to capture complex flow dynamics. Ground Truth & Uncertainty: Unlike standard datasets that rely on external motion capture or LiDAR, this real dataset derives ground truth optical flow via homography-based frame-to-poster registration. Uncertainty Estimation: A unique feature of this dataset is the inclusion of per-pixel uncertainty maps. These are generated via Monte Carlo perturbations of keypoint correspondences, allowing for reliability-aware evaluation metrics that account for registration noise and camera calibration limits. Usage This dataset is compatible with the eWiz library, which provides specific tools for loading the data, visualizing the uncertainty maps, and computing uncertainty-weighted error metrics (AEE and AAE). Library Repository: https://github.com/CIRS-Girona/ewiz Inquiries & Support For any questions regarding the eStonefish-Scenes dataset, or the real-world validation data, or the data generation pipeline, please contact the corresponding author: Jad Mansour | Email: jad.mansour@udg.edu We hope that this work encourages the community to dive deeper into event-based underwater perception, preferably with fewer leaks in the pipeline and more flow in the right direction!