This dataset contains two Wi-Fi databases (one for training and one for test/estimation purposes in indoor positioning applications), collected in a crowdsourced mode (i.e., via 21 different devices and different users), together with a benchmarking utility software (in Matlab and Python) to illustrate various algorithms of indoor positioning based solely on WiFi information (MAC addresses and RSS values). The data was collected in a 4-floor university building in Tampere, Finland, during Jan-Aug 2017 and it comprises 687 training fingerprints and 3951 test or estimation fingerprints. 13.10.2017: Version 2 uploaded; the revised version contains improved readme files and improved Python SW. The dataset and/or the associated software are to be cited as follows: E.S. Lohan, J. Torres-Sospedra, P. Richter, H. Leppäkoski, J. Huerta, A. Cramariuc, “Crowdsourced WiFi-fingerprinting database and benchmark software for indoor positioning”, Zenodo repository, DOI 10.5281/zenodo.889798