The estimation of economic flood losses is a crucial part when decisions about flood defense are taken on the basis of cost-benefit-analyses. However, the field of flood loss modeling has notgained much attention so far. Therefore, improved and scientifically sound loss models are a fundamental step towards a cost-effective flood management. In the aftermath of floods in August 2002, August 2005 and April 2006 extensive datasets about flood losses were collected at affected properties in Germany. Together with other information these datasets were used within the project “Methods for the Evaluation of Direct and Indirect flood losses” (MEDIS) to derive multi-factorial loss models for the residential, commercial, agricultural and public sector. The aim was to improve the hitherto existing methods for flood loss estimation. Therefore, model development was accomplished by a thorough validation of the new loss models and by comparing their performance to currently used stage-damage-curves. To provide a consistent flood loss data base in the future the MEDIS project has also dealt with improvements of the collection of flood loss data. Therefore, a manual for harmonized loss data collection procedures has been developed. With this, a consistent basis for the development and continuous updating of loss models can be achieved.