Arrest representation by police district

Published July 2026 · Data window: MPD adult arrests, 2013–2025 (pooled)

Across thirteen years of MPD arrest records, in the 2nd District, where 8% of residents are Black, Black people were 65% of everyone arrested, 8.1 times their share of the neighborhood. In Districts 6 and 7, arrests of Black people were roughly in line with who lives there. The same shape as the stops data, over a longer window and with heavier consequences.

2.06×

Citywide, Black people were 89.3% of known-race arrests (275,298 of 308,197) against a 43.3% share of DC's population.

Thin base. District 2 is 8.0% Black by population, so its multiplier rests on a small population denominator. The underlying arrest counts are not thin (15,105 Black arrests), but treat the multiplier as indicative, not precise.

Each district shaded by its local arrest index: magnitude against its own population, not a ranking. Click a district for its stop-representation page.

Districts are listed in numeric order. This is not a ranking: each district is compared to its own local baseline, never placed on a most-to-least list.

District	Local Black pop	Black arrest share	Local index	Black arrests (N)
District 1 (1D)	25.0%	88.6%	3.54×	41,664
District 2 (2D)	8.0%	64.5%	8.08×	15,105
District 3 (3D)	22.0%	82.5%	3.75×	31,074
District 4 (4D)	48.6%	79.8%	1.64×	24,443
District 5 (5D)	56.7%	92.8%	1.64×	47,432
District 6 (6D)	88.4%	97.5%	1.10×	57,766
District 7 (7D)	86.5%	98.2%	1.14×	55,939

Local index = Black arrest share ÷ that district's own Black population share.

How this number is built (and where it's soft)

Arrests measure enforcement, not crime. An arrest record says who MPD arrested and where, not who offended. That is exactly why it belongs on an accountability site, and exactly why it cannot be read as an offense rate.
Race definition. RACE = 'BLACK' (the separate Hispanic flag is not applied). MPD records a separate Hispanic/ethnicity flag; it is preserved in the data but not applied to this index.
Exclusions (reported, not dropped quietly). Arrests with race recorded as unknown (two source labels, UNK and UNKNOWN: 11,944, 3.73% of all arrests) are excluded from the known-race denominator. Arrests with no mappable district (2,256, 0.7%) are excluded from per-district figures. The figures on this page rest on 95.59% of all arrests.
Baseline vintage. District populations are ACS 2019-2023 (5-year) applied to arrests from 2013 onward, so the early years lean on a denominator measured later. Population shares move slowly, but treat early-year local indices as approximate.
Grain. One row per arrest (a person arrested twice counts twice). Counts are arrest-records, paired against ACS person population.
Geography. The arrest district (where the arrest occurred), the same seven-district grain as the stops analysis. PSA-level detail exists in the data and is future work.