Numbers: How many have been taken?

Last update: April 9, 2021

You have probably heard the figure "one million" thrown around in discussions of mass internment and reeducation in Xinjiang. It's a rough figure; only the Chinese government and Communist Party know the real numbers. China has refused to publish any official figures or statistics. The rest of us are left guessing.

Concern trolling about inaccurate numbers of detainees without simultaneous acknowledgment of the fact that there is a reason we don't have accurate numbers is gaslighting. These estimates people make are inherently imperfect and imprecise, of course; but as I will explain below, they are at least plausible attempts at quantification. The response of the Chinese government is "nuh-uh." That's it. That is literally all we have to work with.

As adults with faculties of reason, we have to make judgment calls about not only how logical an estimate may sound, but also evaluate them against the limited amount of information on the extent of the camps. This is hard. Jessica Batke summarizes it well in her overview of various detention figures from January 2019, a piece I highly recommend:


Much reporting on the estimates of those detained in Xinjiang stresses that the “one million” number is “credible.” Yet it is important to distinguish between credibility and precision. Much of the information that international observers have used to make detention estimates is credible in that it comes from local sources, many of whom are in official positions that allow them access to such information through the course of their work, and who take great personal risk to communicate this information to the international community. Yet it is observers’ inability to conduct any sort of independent verification that prevents these credible estimates from being more precise. This is mainly due to China’s stringent information controls and its restrictions on foreigners’ access to the region—let alone to the camps themselves.

So, here are some numbers. They range from 0 to a few million. What you think of them is up to you—there is certainly plenty to debate.

Some percentages

Elsewhere on this blog, I have posted translations of various texts by cadres and security officers dispatched to Xinjiang, most of which have been collected by Professor Timothy Grose and archived here. These posts provide further evidence that detentions are widespread, far exceeding the capacity of incarceration facilities in 2017. We cannot generalize from the village level to all of Xinjiang, of course, but combined with other forms of evidence and other estimates, these individual data points tell us that there are at the very least some areas of Xinjiang where a staggering number of people are uprooted from their communities.

  • 11%—A work report partially recovered from an online, for-profit report writer website stated 11% of a certain village's population had been detained, imprisoned, and/or sent for reeducation by the end of 2017.
  • 16.5%—In August 2018, a cadre in a village in Maikit County reported that 103 out of 624 villagers (16.5%) had been detained or otherwise taken away in the ongoing crackdown.
  • 32%—A teacher from a school in Inner Mongolia that ran a donation drive for a school in Qarasaz, Xinjiang stated that she learned 32% (1,744) of the elementary schoolers in the township of Qarasaz have had at least one parent taken away.
  • 365%—A senior police officer in this post stated that he was temporarily assigned to a detention center in spring 2017 operating at approximately 365% capacity.

Note that not all social media posts give such high numbers, and most are hardly so specific. In one post from April 2018, cadres say they visited 189 households, of which 6 six completely empty due to the residents taken for reeducation. Many more households presumably had fewer than every single person missing, but we can't know exactly how many. One post from July 2017 indicate only about 4% (18 of 434) of households had someone missing. Another from May (about two months after the beginning of the current campaign) states only 11 out of a population of 1038 Uyghurs—less than .1%—had been taken at that point. Clearly, there is significant variance across time and place.

11,000 to 130,000

Estimated detention capacity of Dabancheng camp circa October 2018 by architecture firm

BBC was one of many organizations that published satellite imagery analysis of Xinjiang. For analysis of one particularly large camp, BBC recruited GMV, a Spanish technology firm. GMV does include defense contracting as one of its areas of business, it should be noted.

Of the 100-plus facilities BBC identified, GMV categorized 44 as highly or very highly likely to be security facilities. It analyzed the growth of these 44 facilities (some of which are shown in the BBC article).


GMV calculates that, from this set of 44 sites alone, the surface area of secure facilities in Xinjiang has expanded by some 440 hectares since 2003. This measurement refers to the whole site within the external security walls, not just the buildings. But 440 hectares represents a lot of additional space. For context, a 14-hectare site within the city of Los Angeles – containing the Twin Towers Correctional Facility and the Men's Central Jail - holds a combined total of almost 7,000 prisoners.

BBC then contacted an Australian firm, Guymer Bailey Architects, which focused on the camp in Dabancheng. Much is unknown about the facility, but Guymer Bailey concluded at the very least it could hold 11,000 people (Riker's Island in New York, the USA's largest prison, can hold 10,000). This number assumes detainees are held in individual cells; if they are held in dormitory-style shared rooms, that number could be theoretically as high as 130,000, though a number this seems less likely due to the massive logistical constraints that would present.


The number of people placed into two purpose-built detention camps in 2017 in Yarkand County (with a population of roughly 900,000 Uyghurs), according to leaked documents published by the New York Times. The ASPI database lists nine Yarkant reeducation facilities in total (not including prisons). Per the Times:


Mr. Wang [Yongzhi] set about beefing up security in Yarkand but he also pushed economic development to address ethnic discontent. And he sought to soften the party’s religious policies, declaring that there was nothing wrong with having a Quran at home and encouraging party officials to read it to better understand Uighur traditions. When the mass detentions began, Mr. Wang did as he was told at first and appeared to embrace the task with zeal. He built two sprawling new detention facilities, including one as big as 50 basketball courts, and herded 20,000 people into them. But privately, Mr. Wang had misgivings, according to the confession that he later signed, which would have been carefully vetted by the party.

(See my post about ASPI satellite imagery here, and about the authenticity of the Times leaks here.)

This number is particularly notable when we consider that there are other detention facilities there besides those two, and that more detentions may have occurred in 2018.


230,000 people in Xinjiang formally imprisoned from 2017-2018, a several-fold increase over previous years. This number is derived from data provided by the Xinjiang government for 2018 and combined with identical analysis for 2017 by the Network for Chinese Human Rights Defenders. CHRD is strongly anti-CCP, but regardless of this bias, its Chinese sourcing is meticulous and easily accessible in the report. Recall that formal imprisonment differs from internment/reeducation.

The New York Times explains the significance of this figure in the context of previous years. Note the massive spike starting in 2016-2017, when Chen Quanguo arrived in Xinjiang as Party Secretary and began implementing the ongoing mass internment and reeducation campaign. Documents cited by the campaign include official Xinjiang government judicial work reports. The 2018 report (delivered January 2019) states 133,198 people were sentenced, with only 22 declared innocent.


Other things to note:

  • Sentences are harsher: "A greater share of defendants in Xinjiang were sentenced to prison terms of 5 years or longer in 2017 compared to those in other areas of China." (87% versus 14%)
  • These sentences are disproportionately high for Xinjiang's population: "Xinjiang accounted for less than 2% of China's population but 21% of arrests in 2017, a sharp increase in its share of arrests from a decade ago."
  • Caveats: "Xinjiang, like other parts of China, does not disclose how many people are in prison, and the regional government did not answer faxed questions about incarceration and the legal statistics. Not all the people imprisoned in Xinjiang are from Muslim minorities, and not all charges are baseless." Still, "Arrests and indictments in areas run by the Xinjiang Production and Construction Corps — a quasi-military administration overseeing areas with an 85 percent Han population — rose modestly or remained flat in 2017."


The estimated number of people detained by early 2018 by Zenz

This figure comes from the very tail end Zenz 2018, which I cover in detail here. It represents a very separate argument from the bulk of that paper, which focuses on tracking camps through publicly available construction bids, and ultimately relies on leaked, unreproducible data. Zenz is highly conservative and works for the Victims of Communism Foundation. Your level of trust in him might be low, depending on your political persuasion (mine certainly was, which is what motivated me to start this whole project to begin with).

First, the data: a spreadsheet given by an Istanbul-based Uyghur organization to Japanese researcher Mizutani Naoko, which which she detailed in a March 2018 article for Newsweek Japan (link, in Japanese). Mizutani is a well-known scholar in Uyghur studies in Japan and lecturer at Chuo University in Tokyo and associates with Uyghur exile political figures that the Chinese government considers illegal separatists. She has travelled extensively to Xinjiang, but in 2010 she was denied entry to China by authorities. (She quite exasperatedly expressed her disappointment here.)

Zenz does some basic math based off of the data Mizutani produces in her article.


According to this source, the 27 counties in Khotän, Kashgar and Aqsu Prefectures, with a combined Uyghur, Kazakh and Kyrgyz population share of 90.2%, had 693,273 detainees in mid-February 2018. This would represent 12.3% of their Muslim population aged 20–79 (of 4.45 million) [...] Assuming a Muslim adult internment rate of 10% for cities with a majority-Muslim population and of 5% for cities where the Muslim population is less than 50%, Xinjiang’s total reeducation internment figure may be estimated at just over one million (approx. 1,060,000). This would suggest an overall internment rate of Uyghurs and Kazakhs (aged 20–79) of up to 11.5% (12.3% for Aqsu, Khotän and Kashgar, and 10.2% for other regions). The accuracy of this estimate is of course predicated on the supposed validity of the stated sources.

Again, this ultimately relies on information neither you nor I can independently verify. Such is the nature of leaks. As we shall see, though, various other forms of evidence help established that this is at least a plausible figure. As noted above, the Chinese government certainly has this data, but it is choosing not keep it secret; thus, we are left to educated guesses.

About 1.1 million

The estimated number of people detained based on a small-sample survey of exiles

This figure was reported by the Network of Chinese Human Rights Defenders (CHRD) along with the Equal Rights Initiative in August 2018, a month before Zenz' article (though almost certainly while said article was in peer-review). The Equal Rights Initiative is an Alabama-based nonprofit that opposes mass incarceration and advocates for prisoners in the US criminal justice system. The figure was derived from a limited set of interviews:


Between July 2017 and June 2018, we interviewed dozens of ethnic Uyghur villagers in several counties in Kashgar Prefecture (喀什地区), of the Southern Xinjiang (南疆) sub-region and others. All the interviewees spoke on condition of anonymity out of fear of retaliation by Chinese authorities. [...]

CHRD then presents this table based on the approximated figures from their interviewees, each who lived in different Kashgar villages:


The margin of error here is extremely high, as should be obvious; this methodology is ultimate more like an SAT math word problem than rigorous academic analysis. Such criticism cannot take place in a vacuum and pretend to be intellectually honest, though, ignoring the totality of evidence complementing the estimate. Bottom line: the fact that this estimate accords with other data indicates it is a likely reliable ballpark figure. We cannot rely on the NCHRD estimate alone, but absent compelling evidence otherwise—keeping in mind, again, that the Chinese government refuses to release any numbers of its own—there is no reason to discount it out of hand. It is a flawed but workable piece of evidence that must be understood in the light of other evidence as well.

3 million

Randall Schriver, a rather bombastic United States Department of Defense official, asserted this number to a reporter in May 2019. I have not seen any substantive evidence or analysis suggesting this to be the case and thus I do not consider it very credible. Frankly, it is only worth mentioning because it is occasionally parroted by others, though most media outlets stick to the "about 1 million" figure.

This also is notably higher than the 800,000 to 2,000,000 figure the US Department of State cited for 2018.