I attempt to disentangle a problem of causality between institutional quality and interpersonal trust using evidence from a natural experiment: mid-2000s institutional reforms in the post-Soviet nation of Georgia. The reforms following the 2003 Rose Revolution were swift and extensive, aiming mostly at combating corruption and organized crime, improving law enforcement and economic liberalization. At the same time, the neighboring nations of Armenia and Azerbaijan, both former Soviet republics with cultural and economic background similar to the Georgian one, experienced no such change, thus becoming credible counterfactuals to Georgia. To reduce unobservable heterogeneity between Georgia as a treatment group, and Armenia and Azerbaijanis a control group, I exploit the fact that republics borders during the Soviet era did not always reflected the settlement patterns of ethnic groups, thus creating a number of minorities separated from their ethnic kins by arbitrary borders that were internal within the USSR but have become international after the independence. In this particular case, Georgia has several districts with predominantly Armenian and Azeri population spanning along its border with Armenia and Azerbaijan. Comparing people of the same ethnic group on both sides of the border allows concentrating on differences in governance and formal institutions and to diminish possible confounding effect of culture-related heterogeneity. Applying regression discontinuity design to the data from Life in Transition and Caucasian Barometer surveys, I find that Armenian and Azeri residents of Georgia have greater level of interpersonal trust than their counterparts in Armenia and Azerbaijan. Perceptions of corruption and rule of law are likely channels of influence.