Assessing the infection fatality rate (IFR) of SARS-CoV-2 is one of the most controversial issues during the pandemic. Due to asymptomatic or mild courses of COVID-19, many infections remain undetected. Reported case fatality rates - COVID-19-associated deaths divided by number of detected infections - are therefore poor estimates of the IFR. Endogenous changes of the population at risk of a SARS-CoV-2 infection, changing test practices and an improved understanding of the pathogenesis of COVID-19 further exacerbate the estimation of the IFR. Here, we propose a strategy to estimate the IFR of SARS-CoV-2 in Germany that combines official data on reported cases and fatalities supplied by the Robert Koch Institute (RKI) with data from seroepidemiological studies in two infection hotspots, the Austrian town Ischgl and the German municipality Gangelt, respectively. For this purpose, we use the law of total probability to derive an approximate formula for the IFR that is based on a set of assumptions regarding data quality and test specificity and sensitivity. The resulting estimate of the IFR in Germany of 0.83% (95% CI: [0.69%; 0.98%]) that is based on a combination of the RKI and Ischgl data is notably higher than the IFR estimate reported in the Gangelt study (0.36% [0.29%; 0.45%]). It is closer to the consolidated estimate based on a meta-analysis (0.68% [0.53%; 0.82%]), where the difference can be explained by Germany's disadvantageous age structure. As a result of virus mutations, vaccination strategies, and improved therapy, a re-estimation of the IFR will eventually be mandated; the proposed method is able to account for such developments.