The author argues that the majority of research on police violence in Canada is anecdotal, and the available statistics are primarily situational. He develops an original database, based on newspaper sources, to introduce the problem of police violence in Canada. The creation of the dataset follows the methodology used in Ross (1988; 1992a; 1993; 1994), which involves the development of a detailed chronology of the violent event, the coding of these events on a series of relevant variables, and the running of frequency distributions. The author constructs a chronology based on newspaper articles that referred to acts of municipal police violence in Canada. Eighteen variables are coded. They pertain to the characteristics of each event: year, month, and day of the event; Province where the incident occurred; the police force responsible for the real or alleged violent; most severe type of violence; race or ethnicity of victim; race or ethnicity of the police officer; victim's age; police officer's age; genders of victim and police officer; number of victims; number of police officers; and the most severe outcome or sanction meted out to the officer. Following the description of the data-collection methodology, the chapter discusses the control of police violence. It addresses types of control, the control of police organizations, and the control of police violence and how to improve it. 11 tables, 31 notes, and 94 references