This dataset contains data from the Slide-Study data set used in the paper: [1] C. Mercan, M. Balkenhol, R. Salgado, M. Sherman, P. Vielh, W. Vreuls, A. Polonia, H. M. Horlings, W. Weichert, J. M. Carter, P. Bult, M. Christgen, C. Denkert, K. van de Vijver, J.-M Bokhorst, J. van der Laak, F. Ciompi, Deep learning for fully-automated nuclear pleomorphism scoring in breast cancer. NPJ Breast Cancer, 2022. The dataset consists of n=118 digital pathology whole-slide images (WSI) of breast cancer surgical resections, stained with hematoxylin and eosin (H&E) at Radboud University Medical Centers, Nijmegen (The Netherlands). The WSIs were scanned with a 3DHistech P1000 scanners at 0.25 um/px spacing, originally stored in MRXS file format. However, the WSIs made available here have been converted to TIFF format with a maximum spacing of 0.5 um/px. This was done to make slides broadly accessible (since MRXS files are sometimes not compatible with some digital pathology viewers or APIs), and with the same spacing used in the prediction of the pleomorphism score in the NPJ breast cancer paper. Note that we are solely releasing the Slide-Study test set used in [1]. Together with the data, we have released a web-based evaluation platform via the grand-challenge.org platform, which can be found at this link: https://breastpleomorphism.grand-challenge.org/. In this way, researchers can download the WSI from Zenodo, process them with their algorithm to predict a single pleomorphism score for each slide, compile the predictions as indicated on the grand-challenge.org page, and submit them, to compare the results with the ones presented in the paper and with the opinion of a panel of four pathologists involved in the study. The data is released under CC BY-NC 4.0 license.