This article investigates how the Indonesian state organised the killing of approx. 100,000 communists and alleged communists in Central Java in 1965. It presents the argument that even though state institutions unleashed the killings and perpetrated much of the violence, the state’s control over this violence was limited. In particular, decisions by state institutions as to who would be targeted by the violence at the individual level were considerably influenced by civilian actors. Six theses develop this argument by reconstructing these events. They highlight the fact that the Indonesian army faced capacity constraints (thesis 1) and relied on improvisation (2). The army detained many of the victims in improvised facilities prior to their deaths (3). In these installations, the army’s capacity to identify and select those of the detainees it wished to execute was constrained by a lack of reliable men among their forces. Chaotic conditions in the detention facilities put further limits on the state’s capacity to select people for execution. To counter these effects, auditing and investigation teams were put into place to carry out these selections (4). In doing so, they had to rely on information from their victims’ social environments (5), which identified candidates for detention and supplied details that helped the selection teams decide what to do with detainees (6). This information was supplied voluntarily, often as a result of personal initiative.