The death penalty remains one of the most controversial issues of criminal punishment not only because of racial/ethnic disparities, wrongful convictions, and inadequate defense representation but also because of the potential for geographic arbitrariness. The key empirical proposition embedded in the concept of geographic arbitrariness is that localities react to legally and procedurally equal capital cases in much different ways. This study assesses whether this key proposition is observed. We use data on 880 capital murder cases, nested in 18 counties in Pennsylvania from 2000 to 2010, to examine between-county differences in (a) prosecutors' filing of specific aggravating circumstances, (b) prosecutors' filings to seek the death penalty, (c) prosecutors' decisions to retract a death filing, and (d) court (judges' or juries') decisions to impose the death penalty. First, we found that counties differed substantially in their decisions connected to the same statutory aggravators and murder characteristics. Second, propensity score weighting models demonstrated meaningful and sometimes huge differences between specific counties in prosecutorial decisions to file to seek the death penalty, and to retract those filings. We also found meaningful differences in the likelihood of defendants being sentenced to death across counties, although differences were not as large as those for the prosecutorial decisions. Little evidence exists that county partisan