Between 2009 and 2015, the communities located in the Lower Sava valley experienced five flood events. The floodevents occurred due to continuous rain in central and north-eastern parts of Slovenia. The least threatening of the floodevents included increased water level of the Sava and the Krka rivers, which isolated only few houses from the rest of thecommunity. The most devastating event caused several roadblocks, flooding the entire areas of the communities locatedclose to the two rivers. The data on the rivers' flow rates and levels during the flood events was obtained from the SlovenianEnvironment Agency database, and data on the severity of the flood events from the Administration for Civil Protection andDisaster Relief database. We merged both types of data in a single database and created a timeline of the events with riverdynamics for every event. Based on past events, the communities have learned how to react and protect any endangeredproperty. The communities near the Sava and the Krka in the Lover Sava valley date back to the times before the Franciscancadastre. Floods occurred several times in the past, but the respective communities learned their first significant lesson onlyin 2010 when they were affected by a flood of historic proportions. Several types of tacit knowledge emerged during thatevent and the events that followed almost every year since. We identified a new knowledge base concerning when, to whatextent, and how to organize the protection of threatened households. To be able to create a community-learning model, weconducted semi-structured interviews with people from the households threatened by the flood events after 2009. Thelearning model, supported by a timeline of the events, revealed which event affected the learning process and how. Basedon the emerged knowledge, communities not only changed their own behaviour but also influenced the response processof public services. The influences manifested themselves as two type of information delivered to public services. The firsttype provided public services with new insight into endangered areas that would otherwise remain undisclosed, along withthe need for distress assistance. The second type of information provided public services with an updated overview of thelocal water level situation, not covered by the official reports. Based on the community's informal information source, publicservices were able to adjust their on-field response process in order to support the endangered communities. The data oninformation exchange was taken from the database of the national Administration for Civil Protection and Disaster Reliefdatabase, and local Civil Protection Command logbook. In the final learning model, we merged the data on response processmodification with the timeline of the events and the community learning process. We used different statistical methods todiscover which community performed best as a learner, influenced public services the most, and why. We also determinedsome learning rules, typical for the chosen communities in the Lower Sava valley and defined behavioural and learningcorrelations among different households in the Lower Sava valley.