The article presents a drone sensory database collected during flights with different types of propeller failures. Measurements from four accelerometers and four gyroscopes were collected during 20 flights with two types of faults occurring in different configurations in one, two, three or four rotors. The paper shows the architecture of the system and the procedure for acquiring and processing the data. Raw sensor outputs, pretreated data, and digitally processed signals were provided in a publicly available repository, the structure and purpose of which are discussed in the paper. The applicability and potential use of the shared data for other research are indicated. The provided repository should be helpful in developing methods for detecting and classifying faults in actuators of unmanned aerial vehicles (UAVs). It will be particularly useful for researchers working on data-driven methods. The default purpose of the dataset is to train artificial intelligence models that require large amounts of data.