Divorce in King County, 2012–2022

Methodology and limits

This analysis is built from a manual records pull from King County Superior Court covering documents filed between January 2012 and December 2022. The data has three defects that shape everything above. We state them here because they are not obvious from the file, and because two of them make the most tempting charts false.

The 90-day wall is a floor, not a countdown from filing

RCW 26.09.030 requires ninety days to elapse "since the petition was filed and from the date when service of summons was made upon the respondent or the first publication of summons was made" before the court may enter a final dissolution decree. The clock runs from the later of filing and service, not from filing alone, and what it gates is entry of the decree. Chart 1 above measures days from filing only, so day 90 there is a floor under the true statutory minimum — the wall sitting this sharply at day 90–91 implies service was made at or near the moment of filing in the overwhelming majority of cases, which is itself a finding, not an artifact.

The waiting period is commonly called a "cooling-off period," but the statute does not use that phrase, and Washington's digitised legislative record does not reach the 1973 act that created it, so no recorded legislative purpose is available without a State Archives request. The closest citable authority is dictum. In In re Marriage of Wilson, 117 Wn. App. 40, 68 P.3d 1121 (Div. III 2003), the majority — while rejecting an unrelated argument that a similar waiting period applied to parenting-plan agreements — wrote that "a cooling off period may, for policy reasons, be required before dissolving the marital status." That sentence is subjunctive, is dictum on a question the court was not deciding, and originates as the opposing party's own characterization; it supports only that a Washington appellate court has assumed, without deciding, that such a rationale may underlie the waiting period — not that "Washington courts have held" a cooling-off purpose. The same opinion states the mechanics without hedging: "under RCW 26.09.030 the parties in dissolution proceedings must wait 90 days before entering a final dissolution decree after filing and service of the summons and petition."

Filing volume cannot be compared across 2021

King County began automatically issuing a Temporary Order Setting Financial Restraints on every dissolution around 2021. That document appears on 0.0% of cases filed 2012–2017, 4.8% in 2020, then 99.1% in 2021 and 100.0% in 2022. The consequence is that 2021–2022 captures essentially every dissolution filed, while 2012–2020 captures only cases that generated a support, restraining, or relocation document — roughly 44% of actual filings. A chart of total filings per year would show a doubling in 2021 that did not happen. We do not publish one. Every year-over-year figure above is restricted to cases carrying at least one non-restraints document, which is defined identically in all eleven years.

The domestic-violence flag was abandoned in 2018

The share of cases flagged for domestic violence runs 28.4% in 2012, declines to 11.3% by 2017, drops to 3.2% in 2018, and reads exactly 0.0% from 2019 onward. Domestic violence did not end in 2019; the field stopped being filled in. We publish no domestic-violence trend line and restrict all domestic-violence figures to 2012–2017.

Recent cases have not finished yet

The extract ends on 11 August 2023. At that date, 19.2% of cases filed in 2022 were still open. Their eventual duration is unknown, so a simple median of completed cases would measure only the fast ones and make recent years look artificially quick. We publish no median time-to-resolution for any year this affects. Duration over time is reported as a Kaplan–Meier survival curve, which treats an unresolved case as censored — known to have lasted at least this long — rather than discarding it. One median does appear on this page: the 2012–2017 domestic-violence comparison below. Censoring cannot distort that one — only 2 of those 13,926 cases (0.01%) were still open at the pull.

Two claims we can now make that the flag alone could not

The domestic-violence flag is unusable after 2017, but restraining-order documents are counted, not flagged, and they span all eleven years. They are what let us show the flag was broken rather than the violence ending. They are not a measure of domestic violence itself — many cases involve abuse without a protective order ever being filed — so we report them as what they are: a count of protective-order proceedings.

Duration by disposition route is restricted to 2012–2019 filings for the same reason the survival curves exist: you only learn how a case ended when it ends, so including years that are still running would make the slow routes disappear and flatter the fast ones.

What is in the file

76,184 documents across 30,274 cases. Twenty-two rows were excluded: they are entirely empty spreadsheet padding, with no case number, filing date, or any other field filled in, and are listed as-is in the published dropped_rows.csv log. A further 1,622 rows were duplicates: the source is a join of documents against dispositions, so a case that was resolved and later reopened repeats each of its document rows once per disposition. Two hundred and twenty cases carry more than one disposition; each is reported at its earliest one — the decree that ended the marriage — with the reopening preserved in a DispositionCount column. The extract contains 30 document types, all of them child-support, restraining-order, or parenting-plan and relocation filings; it contains no petitions, summonses, or decrees. Litigant names are omitted from every published file. Case numbers are retained so any row can be checked against the court's own public index.

Get the data

• cases.csv — one row per case, with the population and domestic-violence-validity flags used above
• documents.csv — one row per document filed
• dropped_rows.csv — the 22 excluded rows, all entirely empty
• figures/ — the exact numbers behind each chart, one file per figure

Cite this

Published under CC BY 4.0. Source: King County Superior Court records, obtained by manual records request.