Despite continued cases of sudden in-custody death in subjects who are restrained, little is understood about the incidence of sudden death, its underlying pathophysiology, or its actual relationship to subject positioning. This paper reports data from 4828 consecutive use of force events (August 2006-March 2013 in 7 Canadian police agencies in 4 cities including Eastern and Western Canada. Methods: Human subjects committee approval was obtained in each city with approval for enrolment of subjects without consent. Consecutive subjects aged >18 years who were involved in a police use of force event were included regardless of the cause or outcome. Officers prospectively documented the final resting position of the subject, whether there was knowledge or suspicion that the subject was intoxicated with alcohol and/or drugs, suffering from emotional distress or any combination of those, the number and nature of a standardized list of features of excited delirium present, and the use of all force modalities alone or in combination. Our outcome of interest was sudden, unexpected in-custody death. Confidence intervals for differences were considered significant if the 95% confidence interval for the difference did not include zero. Assuming an alpha error of 5%, a beta error of 20%, a sample size of 1945 subjects per group gives our study 80% power to detect a difference of 0.5% in sudden death between the prone and not prone position. Results: During the study there were over 3.25 million consecutive police-public interactions; use of force occurred in 4,828 subjects (0.1% of police public interactions; 95% CI = 0.1%, 0.1%). 90% of subjects had complete information on positioning; none of the subjects without documentation of positioning died. Subjects were usually male (87.5%); median age 32 years. Subjects were abnormal with 81.5% of subjects documented to have one or more of alcohol and/or drug intoxication, and/or psychiatric/emotional distress at the scene. Significantly more subjects remained in a non-prone vs. prone position; yet over 2000 subjects remained prone. Although 5.3% more subjects with any assessed comorbidity were in the "not-prone" position, over 1500 with any assessed comorbidity were prone. Significantly more individuals with >3 features of excited delirium were not-prone while significantly more subjects with drug intoxication alone were in prone. There was no difference in CEW deployment in any mode between the positions. One individual died suddenly and unexpectedly; the individual was in the not prone position and exhibited all 10 features of excited delirium. No subject died in the prone position. In a worst case scenario our confidence intervals indicate, with a high degree of precision, that 99.8% of subjects would be expected to survive being in either the prone or notprone position following police use of force. Conclusions: Prone positioning is common following police-public interactions. In this study no subject died among thousands who remained in the prone position after po...