Most countries collect loss and damage data after disasters for learning purposes and in support of future preventive work. The lack of international standards and sharing principles implies heterogeneous data sets, thus presenting a challenge to the development of indicators intended to assess progress within the UN agreement Sendai Framework for Disaster Risk Reduction 2015-2030 (SFDRR). In this study, data on mortality, affected people and direct economic losses are extracted from two national databases in Sweden for the years 1996-2015. Pre-SFDRR terminology, definitions and different inclusion criteria are used to exemplify and identify challenges when global "proxy" data inquiry clashes with sub-national demands for data quality. Different test methods on how to estimate affected people are used and in comparison with the term 'directly affected people', as proposed in the SFDRR indicator establishment process, it is concluded that methods for more disaggregated data are needed. In a Swedish context, the SFDRR call for a reference period 2005-2015 is found to be a time too short for providing a fair picture of disaster risks within Sweden's borders. The nationally developed strategy in Sweden, as in many other countries, to learn in-depth after each new disaster and use the experience to remedy weaknesses in safety systems, generates solid data supporting the development of SFDRR indicators, but the national benefits and the relevance of statistics from disasters re-occurring on longer time scales are limited.