• EN
  • 中文
08 Mar 2021

Presumed Guilty report and trials of Michaels Spavor and Kovrig

On account of the announced trials of Canadian businessman Michael Spavor, on Friday, March 19 (in Dandong, Liaoning province), and former diplomat Michael Kovrig, on Monday, March 22 (in Beijing), both on charges related to national security and espionage, Safeguard Defenders has brought forth its release of Presumed Guilty. Its presentation of official data on arrests, prosecutions, trials, and convictions in the Chinese criminal justice system will serve as an invaluable background resource to understand the issue of fair trials, judicial standards, and how, statistically, once one is arrested, a conviction at trial is near-guaranteed. Data covering the rule of Xi Jinping, 2013 to 2020, is presented, along with some additional historical data. This includes the brand new data reported to the Chinese National People's Congress (NPC) on March 8, covering the most recent year, 2020. 

Download the report directly from our publications page, or read it in webform below. 


Presumed Guilty

A briefing on data concerning arrests, prosecutions, and trials in China 2013-2020


Table of contents 

Executive Summary


Trials and convictions

Arrests and Prosecutions

Other issues of concern



Executive Summary

This briefing is focused on providing the data that underpins the operation of the Chinese criminal judicial system. It draws on data from China’s highest court and prosecutor’s office, and presents information on the number of arrests requested and approved, the number of prosecutions initiated, the number of trials and the likelihood of such trials ending in convictions.

Since the year 2000, when China already held a very high conviction rate, the rate of convictions has soared. However, the current near-guaranteed rate of conviction did not begin with the reign of Xi Jinping, but rather within the first few years of his predecessors’ term. Since Xi Jinping came to power, the rate has further increased somewhat, varying between 99.94 and 99.96 percent. The more freewheeling’ days of Jiang Zemin’s reign, when conviction rates dropped to a low of 98.97 percent, are long gone.

While increasing the conviction rate, the number of judgments since 2000 also exploded. In 2000 China had only some 650,000 judgments issued, with some 6,617 people declared not guilty. By 2019, the number of judgments issued had reached 1,660,637, an increase of some 250%, yet the number of not guilty verdicts collapsed to a mere 637.

During the period 2013 to 2019, the number of approved arrests saw a marked increase from 879,817 in 2013 to 1,088,490 in 2019. Likewise, the number of prosecutions carried out increased from 1,324,404 in 2013 to 1,818,808 in 2019. The amount of arrestees that are not prosecuted remains very low. By 2019, only 5.36% of prosecutions were dropped for insufficient evidence, a slight increase from the first year of Xi Jinping’s era, but still markedly low (in 2020 reduced to a mere 2.54%). If one includes the only other reason why prosecutions can be dropped, i.e. if they are deemed unnecessary, the total figure of all dropped prosecutions in 2019 stood at 7.94%.

These figures mean that if arrested one stands a more than 92 percent chance of facing trial. At trial one has a 99.96 percent likelihood of being found guilty. This means that out of nearly 1.7 million judgments, only about six hundred were declared not guilty.

This increase in the near guarantee of being sentenced upon arrest is further impacted by the fact that since 2013 a number of alternative methods for investigation and detention have been developed, and greatly expanded in size. These include the use of RSDL, and the use of liuzhi by the non-judicial organ the National Supervision Commission (NSC). In addition, although no data exist for now, there seems to have been a continuation, and possible expansion, of the use of various forms of administrative detentions and punishments. Putting these developments next to the massive increase in trials and the plummeting of those found not guilty, it presents a legal system that has not only expanded its reach but presents also the near certainty that anyone targeted will be found guilty.



China’s legal system is notorious for its high conviction rate. The scope of China’s judicial system has continued to expand as the country modernizes, while control over society, including the judiciary, has tightened since the rise of Xi Jinping. The overall signal under Xi Jinping is one of harshened repression and persecution for anyone who in anyway challenges or is perceived as a threat to the power of the ruling party, the Chinese Communist Party (CCP).

However, conviction rates alone do not provide a full picture of the workings of the Chinese criminal justice system. This report therefore analyzes those conviction rates and their variations over time, supplemented with data on arrests and prosecutions. Unfortunately, no reliable information exists on detentions, i.e. police taking compulsory action against individuals before their possible arrest or setting them free. This is a missing piece, but adding arrests requested, arrests approved and data on prosecutions, whether carried on to trial or dropped, adds to a better understanding of the judicial system.

This briefing limits itself to presenting authoritative data on the above specifics, and does not concern itself with other key human rights concerns within the legal system, such as the prevalence of the use of torture and use of forced confessions, the standards of evidence used at trial, the right of utilizing defense attorneys, and their position of power in courtrooms versus prosecutors, political control over the prosecutor’s office and over judges, etc.

A significant body of work exists to address those issues, including from Safeguard Defenders (available on its website). To make a well-rounded assessment of the Chinese criminal system, those points need be addressed in full. However, the data herein presented is not easily available, and thus presenting this data-centric briefing adds an important resource to study the judicial system.

This report thus shows what happens, statistically, to a suspect that police wants arrested, and the likelihood of prosecution, trial and conviction if such an arrest is approved. The findings are likely to be shocking to many, but also include some possible bright spots.  

The data herein presented comes straight from governing bodies, namely the work reports presented annually to the National People’s Congress (NPC) by the Supreme People’s Court (SPP) and the Supreme People’s Procuratorate (SPP). The data herein is thus based entirely on official statistics.

As there is little transparency with any datasets released by the PRC, the authors recognize that this data might be incomplete, or even intentionally flawed, but it is the only data available. We have tried to verify and complement this data with additional information from the Statistical Yearbook, presented each year by the National Bureau of Statistics, but the datasets are not comparable.  

Note: On March 8 this year the new work reports from the SPC and SPP were released, hence data for 2020 is now available. That data is included in this , but often noted separately after each graph or chart presented. Due to the Covid19 virus, which saw a prolonged partial shutdown of the criminal justice system, rendering the data for 2020 in every respect an outlier. As such, the 2018 and 2019 figures are far more representative of the state of the criminal justice system outside of the Covid19 period.


Trials and convictions

The Chinese conviction rate is often said to be 99%. This is seemingly often made to imply that conviction at Chinese trials is more or less guaranteed. In reality, it is higher than that. But it is important to unpack this figure.

The first data, presented below, looks simply at the outcome of trials (at first instance) from 2013 to 2020, looking at total number of judgments, and how many of those were guilty or not guilty. The year 2019 shows that the conviction rate in public prosecutions (i.e. trials based on cases brought forth to court by the prosecutor’s office) stands slightly above 99.96%. That is, out of ~1,297,000 trials (cases), which lead to ~1,660,000 verdicts (people), only 637 were declared not guilty.


Chart. Conviction rates, 2013 to 2020

Table. Conviction rates, 2013 to 2020

Other datasets, such as information in the Supreme Court’s database on verdicts, excludes a number of different kinds of verdicts, such as those related to national security but also for a great many other cases. We have no way of knowing if such verdicts are included in these numbers or not. The current [2020] 99.95706% is thus the minimum rate of conviction, but it might be higher.


Historical conviction rates

High to sky-high conviction rates are not new in China. However, this historical overview, with data presented by Professor Donald Clarke ifn The China Collection, shows that the current levels of not guilty verdicts are extreme even by more recent PRC standards.

The below figure shows the number of not guilty judgments since 1980.

Chart. Historical not guilty judgments by year 1980-2020

Despite the fact that the number of persons brought to trial has more than doubled between 2000 (646,431) and 2019 (1,660,637), the number of people found not guilty collapsed from 6,617 in 2000 to 637 in 2019. The conviction rate, which in 2000 stood at 98.98%, has since 2017 breached 99.96%.

As for the conviction rate, 1980 to 2020, it has developed as shown below.

Chart. Historical conviction rate by year 1980-2020

The table below presents the historical data on guilty versus not guilty judgments and the resulting conviction rate.

Table. Historical data by year 1980-2020 on judgments, guilty vs not guilty and conviction rate


Prosecutions and trials

The below graph shows the increase in the number of prosecutions, as well as number of trials. The first value, prosecutions, denotes persons, while the number of trials (at first instance) denotes cases –on average each case represents more than one person.

Chart. Number of prosecutions versus number of trials, 2013 to 2020

As annual data for prosecutions launched, for trials held, and for sentences issued are not overlapping (for example, only some 2018 prosecutions led to trials in 2018, and only some 2018 trials led to judgments in 2018), this data merely presents the overall relationship between growth in prosecutions and trials.

Again, the authors of this briefing are treating the 2020 data as an anomaly because of the Covid19 restrictions, which also impacted the criminal justice system, and especially the number of trials held (and people sentenced), as many judicial procedures were either postponed or dropped (see further below).

Table. Number of prosecutions versus trials versus people sentenced, 2013 to 2020

The first conclusion one can draw from this section – trials and convictions - is that the number of prosecutions and trials continues to increase in China. The second is that the conviction rate showed an early peak between the late 1980s and mid-1990s, after which it started decreasing before starting to rise again around the year 2000. The third is that the current high conviction rate was reached around 2010 and has stood at a consistently high level ever since, remaining in a prolonged peak ever since Xi Jinping came to power.


Arrests and prosecutions

The second dataset aims to look, as far as possible, at the number of arrests carried out by the police (Ministry of Public Security, MPS), and specifically at how many arrest requests were made to the prosecutor and how many of those were approved. It also looks at what happens after arrest, whether prosecution leads to trial, and how often prosecution is dropped before trial.

Many critics, and a great many victims, have pointed out that upon arrest, as a suspect, your fate is more or less sealed, and from this early stage until receiving a verdict, a suspect’s only way to affect the outcome of the trial is whether to confess or not. Lawyers and victims alike point to being arrested as being the key moment in any person’s criminal judicial process.

Having this data is thus key to understanding just how certain a person’s conviction is upon his/her arrest, and to see how correct such critics are in the above assessment. In assessing the functioning of the criminal judicial system, the conviction rate alone is not enough, but by adding this data we can get closer to a fuller understanding.


Arrest requests denied by prosecutor

After the expiration of time-limits for initial detention and interrogation - which in Chinese criminal procedure law ranges from two to 37 days (in reality, almost always 37 days)  -, the police must either present a request to arrest the person to the prosecutor’s office or set the suspect free.

The following data shows how many such arrest requests were filed by the police from 2013 to 2020, and how many of those were approved by the prosecutor. This does not include any data on how many people have been detained but are released without the police seeking their arrest, or, for that matter, of police placing the suspect under another form of ‘compulsory measure’ rather than under arrest (for example, ‘Residential Surveillance at a Designated Location’ (RSDL)). The data is very insufficient, but nonetheless important.

The graph below shows the number of arrests approved and arrests denied due to insufficient evidence or prosecutor finding upon review that no crime has been committed.

Chart. Arrests approved versus arrests denied due to insufficient evidence, 2013 to 2020

The likelihood of an arrest being denied by prosecutors has slowly increased from 10.22% in 2013 to 15.19% in 2020, be it with yearly fluctuations as the graph below shows.

Chart. Rate of arrest requests’ denied due to insufficient evidence, 2013 to 2020

This does not paint a complete picture however. There is another situation where a request for arrest is denied, as when such as arrests are deemed to be unnecessary by the prosecutor, but such numbers are often classified. This may be the case when the alleged crime is considered minor, where the suspect is a minor, or where the initial detention, or consideration of prosecution, may itself have achieved its intended outcome.

Data regarding this latter category (unnecessary) has been released only for some years (2013-2015, 2018, 2020). However, we can extrapolate rough estimates[ii] for the missing years (2016-2017, 2019), based on an analysis of insufficient evidence versus unnecessary for the years for which both data sets exists. Based on such extrapolation, we can, in rough estimate at least, get a fuller picture.

Chart. Arrests approved versus arrests denied due to insufficient evidence and unnecessary, 2013 to 2020

Note: The figure for 2020 should best be ignored to see the overall pattern of development, as it saw a significant drop in the carrying out of criminal judicial procedures across the board due to Covid19 restrictions.

While reiterating that data for 2016-2017 and 2019 are extrapolations and not exact data, taken together, we can observe that a fairly significant number of arrests requested by the police are in fact denied by the prosecutor.

Chart. Arrests’ denied due to insufficient evidence and unnecessary (total denials), 2013 to 2020

As no data is released on detentions nor on alternative ‘compulsory measures’, the data remains incomplete, but the above show that the prosecutor’s office does, more frequently than often thought, go against the recommended arrests filed by the police.

Table. Approved versus denied arrest requests, by category, 2013 to 2020



Prosecutions dropped

This data, ranging from 2013 to 2020, shows how often cases are dropped by the prosecutor after arrests has been approved, and before trial. This is, for any suspect, a key instance in their judicial process as if the case is not dropped, as conviction rates shows, they are more or less guaranteed to be found guilty.

Like with arrests denied, there are two types of figures for dropped prosecutions. One set concerns insufficient evidence, or where a prosecutor has concluded no crime has been committed, while the other regards instances where bringing a suspect to trial is deemed unnecessary.

As with the data for arrests denied above, the figures of prosecutions dropped for being deemed unnecessary for the years 2016-2017 and 2019 are extrapolations based on the average from the other years, as data were not presented by the SPP for those years[iii].

Chart. Amount of prosecutions versus prosecutions dropped due to insufficient evidence and unnecessary, 2013 to 2020

Most instances of dropped prosecution stem from insufficient evidence, or where a prosecutor has concluded no crime has been committed. Between 2013 and 2019, the likelihood of a prosecution being dropped for this reason has increased somewhat, although as the graph below shows, with fluctuations year by year.

Note: For 2020, the amount of prosecutions dropped for being unnecessary increased some 500%, as can be seen in the chart above. This is without doubt because the judicial system was overwhelmed by case work after the judicial system was in partial shutdown during the worst restrictions following the initial Covid19 virus outbreak. It is not representative, and will likely return to its pre-2020 rate shortly.

Chart. Rate of prosecutions dropped due to insufficient evidence, 2013 to 2020

If one also includes cases where prosecution has been dropped because deemed not necessary – which can happen for a variety of reasons (i.e. the crime is considered so small it is not worth the resources of the legal system to prosecute, an informal bargain between prosecutor and suspect or suspect’s lawyers been made, or any other reason the prosecutor may have), the chance of having a case dropped stands as follows:

Chart. Rate of prosecutions dropped due to unnecessary, 2013 to 2019

The above data shows a continued growth in suspects set free because the prosecution decided that bringing the suspects to trial was either unnecessary or that there was insufficient evidence to support prosecution. The amount of cases is still relatively small, and the chance of seeing a case dropped remains very limited.

Table. Amount of prosecutions versus prosecutions dropped by category, 2013 to 2020

This section – arrests and prosecutions – allows us to draw two conclusions, one positive, and one negative. The positive one is that the prosecutor does seem to, fairly regularly, deny requests for arrests by police, and the figure is likely higher than many assume. The other conclusion is that if an arrest is indeed approved, and a significant majority are, there is a very high risk of prosecution and trial. In fact, if arrested, the risk of trial stands at above 92%. This supports the conclusion of many lawyers that the little impact a person may have on the outcome of one’s judicial process is before the police chooses to request arrest, or before the prosecutor approves it. Once approved, one’s fate is largely sealed.


Other issues of concern

The above data presents a mixed picture in developments of how certain one’s arrest is if requested by the police, how certain prosecution is after arrest, and how likely one is to be found guilty at trial. Before we move over to conclusions, it is important to note that since 2013 there have been a number of other developments that merit mention.

  • As exposed in a report from SD, the use of ‘Residential Surveillance at a Designated Location’ (RSDL), which has been deemed by relevant UN organs as tantamount to enforced disappearances in many cases, has exploded, from a mere few hundred cases in 2013 – when the system in its current form came into effect – to almost 10,000 per year now. This estimate only regards cases where victims of RSDL are later tried and convicted. A great many RSDL victims are never put on trial, nor in some cases even arrested, and thus not part of these statistics. Placement into RSDL – which can last for six months - precedes arrest.
  • As part of Xi Jinping’s expanded “anti-corruption” campaign, the use of the Chinese Communist Party’s internal police (CCDI, or Central Commission for Discipline Inspection) has increased significantly. In 2018, the system underwent further expansion with the establishment of the National Supervision Commission (NSC). Both – which are the same but use different names depending of use on a party member or non-party member – have an RSDL-like system for secret detention (liuzhi). Unlike RSDL, liuzhi is not even technically part of the judicial system, and precedes not only arrest but any act of the judiciary at all. The most comprehensive review of the system, filed with relevant UN organs in 2019, shows the NSC carrying out some 1,667,000 investigations in 2018, and detaining thousands, incommunicado and at secret locations. Of those investigated by the NSC, an estimated 95.4% are deemed ‘guilty’ by the NSC. Of these - 1,736,000 people in 2018 -, only ~3.7% were handed over to the judicial system for any legal process, such as criminal prosecution.
  • Mass incarceration without due process in the Xinjiang Uyghur Autonomous Region is often carried out under administrative orders, meaning it is not part of the judicial process. The hundreds of thousands, or more, victims of these detentions are not considered part of China’s criminal justice system.
  • Other forms of administrative detention, of which there are several different kinds in China, are likewise not included in any data contained in this report. The most famous of these, the ‘re-education through labour system’ (RTL/RETL) may have been abolished, but many other systems exists. The scope of use of administrative detentions in recent years is unknown.



It is unfortunate that the police (Ministry of Public Security, MPS) does not provide exhaustive work reports to the National People’s Congress in the way the Supreme Court and Supreme Procuratorate do, as data on the ratio between detentions and (requested) arrests is a missing piece for a more complete picture of the criminal justice system. It should also be noted that no information at all is made available with regard to policing actions by the Ministry of State Security (MSS).

In addition, lacking information about the scale of use of mass detentions, often at secret locations, incommunicado, via both the RSDL and liuzhi systems - both of which require no arrest, and China’s continued mass use of ‘administrative’ detentions, leaves any analysis missing key pieces of data.

However, it can be concluded that within the official criminal justice system, the trend of the late 90s of high yet slightly reduced conviction rate was reversed during Hu Jintao’s reign, and has been in a prolonged peak ever since Xi Jinping came to power, with only small variations year by year. 

A worrying part of maintaining such a sky high conviction rate is that it has coincided with a significant and continuous growth in the scope of the judicial system, i.e., the number of arrests, prosecutions and trials. The rate remains steady and very high, all while the system expands and new forms of detentions and punishments have been instituted and also greatly expanded in scope since Xi Jinping came to power. 

The rate of refused arrest requests as well as the rate of dropped prosecutions has improved somewhat over the last six years, but remains low. One small positive is the higher than expected number of denied arrest requests from the police. On the other hand, the rate of dropped prosecutions for insufficient evidence remains very low at just about 5% (except for 2020, where it’s under 3%).

The pessimistic view among many observers, former suspects and Chinese criminal defense lawyers largely holds true: once an arrest is approved, it is almost certain (92.08%) that the suspect will be brought to trial, at which point he or she will have a 99.95 to 99.96 percent chance of being found guilty.

Overall, within the narrow confines of the official criminal justice system, the rate of prosecution and conviction remains similar but in light of new systems for detention and punishment the picture represents a noteworthy deterioration for detention and imprisonment.


[1] Denotes “public prosecutions”, i.e., prosecutions brought forth by the prosecutor’s office. In Chinese law there is a concept of “private prosecution” as well, but it is not relevant for this analysis.

[i] In these work reports large numbers are measured in the thousands, hence the value for large numbers is given in the thousands, unlike that for small numbers, such as non-guilty verdicts, which are given in exactness.

[ii] For the four years for which both data exist, no arrest due to “unnecessary” is at about 73.15% of no arrest due to “insufficient evidence or no crime”. It varies as: 2013: 81.96%, 2014: 73.1%, 2015: 68.42%, and 2018: 69.12%.

[iii] For the four years for which both data exist, released from prosecution due to such being “unnecessary” is at about 40.2% of no prosecution due to “insufficient evidence”. It varies as: 2013: 31.96%, 2014: 44.56%, 2015: 50.76%, and 2018: 33.54%.