Rigorous validation tests against independent data and uncertainty assessment are of principal importance in modelling, but these tests mainly increase the reliability of the model, not necessarily the credibility perceived by water managers and stakeholders. Credibility, as opposed to reliability, is much more connected to, and a result of, the manager-modeller dialogue with its speech and interaction process face-toface at model review sessions with peer reviews. However, the manager-modeller dialogue is also influenced by how water managers and different stakeholders cohere or differ in their framing of the environmental dispute which the modelling project is part of. Thus multiple frames can here influence how credibility and uncertainty issues are perceived by the water managers and stakeholders. We suggest that simply focusing on eliminating uncertainty and testing reliability is not enough. Instead, we recommend a broader focus in the modelling process in order to explicitly take notion of and evaluate the perceived credibility. The modellermanager dialogues must care more about how issues of incomplete knowledge, unpredictability, and multiple knowledge frames are explored and perceived by stakeholders. Experiences from Denmark with reference to the national water resource model (DK-model) and regional model projects are presented and discussed, including tools for facilitating dialogue drawing on social-science and natural-science toolboxes.