At the present day, growing interest is paid to the development of more reliable surveillance systems for maritime situational awareness (MSA). The purpose is to detect, track and classify cooperative and non-cooperative targets. For this reason, great interest is given to low-power/cost High-Frequency Surface-Wave (HFSW) radars as an early-warning tool for over-the-horizon (OTH) applications. However, in HFSW radars there is a trade-off in terms of quality and cost, i.e. the radar system exhibits poor azimuth resolution, high non-linearity, and significant false alarm rate. All these aspects reduce tracking performance if not properly addressed. In this context, the Joint Probabilistic Data Association (JPDA) with the Unscented Kalman Filter (UKF) is proposed. The tracking algorithm behavior is investigated by a comparison between the tracks generated by two HFSW radars, with overlapped fields of view, and Automatic Identification System (AIS) data. A discussion is provided about the possible effectiveness of HFSW radar fusion strategies. Preliminary results from a HFSW Radar experiment are reported and discussed.