Unfortunately, research and development for signature verification is spreading across vast applications without fertile synchronization between the forensic and biometric communities. To take a step towards filling this gap, we propose a new dataset for dynamic signature analysis in forensic settings to stimulate research in this direction. We collected data from 13 healthy adult contributors. Although small, we tried to improve the representativeness of the dataset by recruiting participants with heterogeneous characteristics. Out of 13 participants, 8 were women, only one was left-handed, and 5 were trained handwriting examiners with no previous experience with dynamic signatures. Overall, the participants had an average age of 52 +/- 14 years and an average education level of 15 +/- 4 years. We asked each participant to provide: 20 genuine signatures of an invented person (both a name and a surname of varying complexity were considered), equally split between two acquisition sessions on two different days to account for the within-writer variability ("COMPARATIVE", from K1 to K20); 5 test disguised signatures ("IN VERIFICA", from Q6 to Q10), acquired on another different day; 5 test genuine specimens ("IN VERIFICA", from Q11 to Q15), acquired on another day. Additionally, we asked another group of 13 contributors to make several imitation attempts of the original specimens above and to choose the 5 best imitations, after practice, to use as test imitations ("IN VERIFICA", from Q1 to Q5). Out of 13 imitators, 9 were women, none were left-handed, and 3 were trained handwriting examiners with no previous experience with dynamic signatures. Overall, the imitators had an average age of 40 +/- 12 years and an average education level of 16 +/- 3 years. Considering that for one of the invented names we only have 10 comparative signatures, instead of 20, in total we have collected 12 × 20 + 10 = 250 authentic signatures to use as a reference set and 13 × 15 = 195 test signatures, equally divided between genuine, imitated and disguised signatures. The data were acquired using the Wacom STU-530 signature pad, with 1024 pressure levels and a sampling rate of 200 Hz. We chose this tool mainly because it is one of the most common and professional tablets used daily in POS or customer points. This device is equipped with inductive sensor technology and provides (x, y) spatial coordinates, pressure values and time. Participants wrote using a standard stylus directly on the signature pad. The commonly used Namirial's Firma Certa Forensic software was used, which directly calculates some useful features (displacement = "DISTANZA", velocity = "VELOCITA", acceleration = "ACCELERAZIONE"). The output of this software, which can be used directly by forensic handwriting examiners, contributes to this version of the dataset. Each participant was asked to write while sitting at a table in a comfortable position and speed. The signatories were free to move the tablet according to their needs. Participants were in good health and showed no impairments or symptoms that appeared to affect their writing. The signatures were checked after collection for any technical disturbances. If you think this dataset was useful for your research, please cite: Mazzolini, D., Mignone, P., Pavan, P., Vessio, G., "An easy-to-explain decision support framework for forensic analysis of dynamic signatures", Forensic Science International: Digital Investigation, Elsevier, https://doi.org/10.1016/j.fsidi.2021.301216, 2021 The main purpose of the dataset is to encourage contributions on this intriguing topic. In particular, any contribution to expand the dataset is welcome. Acknowledgement: We would like to express gratitude to the participants in the third edition of the "Dynamic Signature Analysis" training course, organized by the Italian Graphological Association in collaboration with Arigrafmilano, for their help in data curation and for their constructive feedback and comments.