Empirical evidence is critical for democratic policing. Data are paramount for the effective governance of the police. This has become especially clear as police officer-involved homicides have gained national attention in the last half-decade. In the wake of the killing of Michael Brown in 2014, scholars and activists quickly identified and decried the lack of reliable national data on police use of force. Media outlets like The Washington Post intervened, establishing a dataset on all civilians shot and killed by police that has created, for the first time ever, close-to-accurate data on the lethal dangers to civilians from encounters with the police. Yet danger in policing is not just present for civilians that the police encounter. Officers themselves are in danger while carrying out their duties. While many other jobs are statistically more dangerous than policing, including commercial fishing, logging, and roofing, policing is unique in the potential for intentional assault by civilians that officers sometimes face while performing their jobs. Because of this, in the public imagination and in the views of officers themselves, being a police officer is one of the most dangerous jobs in the United States. Many researchers have shown that a preoccupation with danger is a central - if not the central - element of police occupational culture.As officers carry out their various duties as law enforcers, maintainers of order, and social service providers, however, they experience critical incidents from a much wider variety of sources than just intentional violence from civilians. Officers respond to gruesome traffic accidents and crime scenes; they investigate child abuse; they encounter individuals in the worst moments of their lives or who are suffering from severe mental illnesses. All of these encounters create stress that, accumulated over time, can and does place officers at an elevated risk of mental illness (such as PTSD and depression) and can lead to maladaptive behaviors including substance abuse and even suicide.This projects interrogates what is known empirically about the dangers of being an American law enforcement officer and what is knowable, given extant data collection. I explore the ways in which data collection about the dangers of policing is co-constitutive with police culture itself. I explore the questions: What do we know about the dangers of policing, what do we not know, and why? I use a variety of sources in this project to demonstrate the ways in which the data that are available on policing are themselves a cultural product, the result of a social process in which police themselves historically played, and continue to play, a critical role. I explore the history of the International Association of Chiefs of Police (IACP) using archival records of that organization's meetings between 1893 and 1905. I demonstrate the presence, even in the IACP's earliest days, of certain dominant normative orders of policing, including danger and white male hegemony. I explore the IACP's cooperation with and encouragement of the FBI, which started the nation's first data collection on police officer fatalities in 1937. I draw a through line between the IACP, the FBI, and modern data collection which focuses disproportionately on civilian assault to the exclusion of other (and statistically more likely) harms to officers, including mental illness and suicide. I present the databases that track dangers to officers and discuss who runs these datasets, what information they collect, and what knowledge is thus created about the hazards of policing. I suggest, in the spirit of critical data studies, that we cannot take data at face value. Instead, we must continue to understand the ways in which policing data and police culture are co-constitutive.I then present data on police officer injuries from two urban police departments in majority-Black cities in the United States as case studies in what data are available on the dangers of policing. For one department, which I call CPD, I present officer injury data from all causes between 2010 and 2018. In line with earlier research, I show that civilian assaults account for only 11% of all officer injuries. For another department, which I call MPD, I present data from 2015 to 2018 on assaults on officers. I show that, even in a dataset already constrained to civilian assaults, injuries to officers are generally minor; not a single officer from MPD was shot or stabbed during my four-year sample. I use MPD and CPD data to reinforce the understanding that police data are designed to create knowledge about the physical dangers of policing, especially from civilian assault, but are currently incapable of creating reliable knowledge about suicide and mental illness. Data collection practices help to perpetuate the myth that the most dangerous part of policing is the civilians that officers encounter.