This article demonstrates how fine continuous and categorical measures of genocide intensity can be derived from the records of the Rwandan transitional justice system. The data, which include the number of genocide suspects and genocide survivors across 1484 administrative sectors, are highly skewed and con- tain a non-negligible number of outlying observations. After deriving nine proxies of genocide intensity from the data, various sets of these proxies are subjected to skewness-adjusted Robust Principal Component Analysis (ROBPCA), yielding four distinct continuous indices of genocide intensity. The effect of survival bias on these indices is reduced by augmenting the set of genocide proxies subjected to ROBPCA with the distance from an administrative sector to the nearest mass grave. Finally, the administrative sectors are divided into distinct categories of low, moderate and high genocide intensity by means of Local Indicators of Spatial Auto-Correlation (LISA) that allow identifying significant high-high and low-low clusters of genocide intensity.