According to the Federal Bureau of Investigations, U.S. police agencies made 13.7 million arrests in 2009. According to the Bureau of Justice Statistics, U.S. jail admissions totaled 12.8 million in 2009. Hence jail admissions totaled 94% of arrests. Before being admitted to jail, arrested persons can be released if they post bail, if they are granted release on personal recognizance, and for other less common reasons. Among persons arrested for felonies, 32% were released within a day of arrest. Among all persons arrested, the share released within a day is surely much higher. Among persons arrested and released within a day, a large share probably did not spend time in jail. So why does the number of jail admissions amount to such a large fraction of arrests?
The total number of detentions is actually much larger than the number of jail admissions. The jail admissions statistics exclude short-term police lockups:
Excluded from the survey [on which the jail admissions statistic is based] are temporary holding or lockup facilities that are not part of a combined function of a jail jurisdiction, from which inmates are usually transferred within 72 hours and not held beyond arraignment.
U.S. police agencies in 2003 had capacity for overnight or longer detention of about 62,000 adults. The typical maximum detention time in these lockups is 48 hours. Assuming a lockup capacity utilization of 50% and an average detention time of 23 hours implies 12 million admissions to police lockups per year. Some persons admitted to police lockups are then transferred to jail. But since 32% of felony defendants are released within a day after arrest, many persons admitted to police lockups are released without being transferred to jail. A crude guess is that 8 million persons a year are detained in police lockups overnight or longer, and then released without being transferred to jail. Combining jail admissions and police lockup admissions overnight or longer that don’t lead to jail admissions implies a total of about 21 million detentions overnight or longer in 2009. That’s seven million more such detentions than arrests in 2009.
While arrest and detention are serious public actions with large implications for personal liberty, large shares of arrests and detentions apparently occur without clear public specifications and procedures. If 50% of persons arrested are released without a jail or lockup stay, then 14 million detentions not associated with an arrest are occurring. Detentions in temporary (not overnight) facilities would add to this total. FBI arrest statistics include 1.9 million arrests for liquor laws, drunkenness (excluding drunk driving), disorderly conduct, and vagrancy, as well as 3.6 million arrests for “other offenses” not included in the list of 28 enumerated offense categories. The surplus of detentions over arrests doesn’t seem to reflect tight constraints on reasons for arrest.
* * * * *
Data: arrests, detentions and jail admissions in the U.S. (Excel version)
 Calculated from BJS, Pretrail Release of Felony Defendants in State Courts, p. 2,6.
 See Annual Survey of Jails, Technical Documentation, p. 6.
 Detention without arrest can occur for violations of conditions of probation, parole, and pretrial release. However, exits from probation and parole to incarceration without a new sentence amounted to only about 300,000 in 2008. See BJS, Probation and Parole in the United States, 2008, Appendix Tables 3, 4, 12, and 14.
 The survey source on lockups (LEMAS, 2003) distinguishes “temporary holding cell (not for overnight detention)” from “lockup or temporary holding facility separate from jail (for overnight detention)”. Concerning the adult lockup capacity of the latter, the survey explicitly states, “Include only overnight facilities separate from a jail used to hold persons prior to arraignment.” The underlining is included in the quoted source text. See Q1, Q20/Q21 in the LEMAS survey instrument.