Broadcast meteorologists serve a critical and complex role in the communication of weather warnings. Surveys conducted after tornadic events indicate that approximately 85% of respondents receive severe weather warnings from local broadcast meteorologists (Hammer and Schmidlin 2002; Sherman-Morris 2005; Schmidlin 2008). Viewers cite trust of broadcaster advice as a main influence in deciding to seek shelter (Hammer and Schmidlin 2002; Sherman-Morris 2005). Even as cell phones become an increasingly popular way for the public receive warnings, surveys continue to indicate that television is the primary source of tornado warnings nationwide (Silva et al. 2017, 2018). Forecasting a Continuum of Environmental Threats (FACETs) is a next-generation approach to National Weather Service (NWS) watches/warnings that will deliver user-specific, probabilistic hazard information (PHI) for improved decision making. As research with PHI matures, an outstanding question exists about how PHI should relate to the current warning paradigm. Due to their role as intermediaries between NWS forecasters and the publics, two broadcast meteorologists, working in a team environment, were included in each week of the 2018 Hazardous Weather Testbed (HWT) Probabilistic Hazard Information (PHI) project. Research protocols were developed and used to systematically study how broadcast meteorologists interpret, use, and communicate probabilistic information. Broadcast participants performed typical job functions under a simulated television studio environment with chroma key (developed within the Oklahoma Weather Lab). The broadcasters received experimental PHI (tornado, wind/hail and lightning) for displaced realtime events, as well as the warning polygons issued by the NWS during the events. Broadcasters used both on-air and social media outlets to communicate PHI to their audiences. Researchers concentrated on communication challenges and decision points of interest, including when to run crawls, post to social media, interrupt commercials, and interrupt programming. Researchers also investigated the interplay between the PHI plume and the traditional warning polygon, and whether they should be intrinsically connected. Other challenges included the visual display of PHI, including color gradients and layering with other data. Results indicate the continued need for a warning threshold or trigger, however, participants strongly desire the inclusion of probabilistic information in tandem with the warning polygon. Additional results will be shown concerning how participants chose to communicate PHI with multiple layers of weather information.