A book on: Government Mass Murder (“Democide”)


An Academic study: “Death By Government” by R.J. Rummel
(Democides that occurred up to 1987)

Conclusions of the study:
The problem is Power.
The solution is democracy.
The course of action is to foster freedom.


By R.J. Rummel

New Brunswick, N.J.:
Transaction Publishers, 1994.
Truth will come to light; murder cannot be hid long
—-Shakespeare, Merchant of Venice

Figures and Tables
Forward (by Irving Louis Horowitz)
1. 169,202,000 Murdered: Summary and Conclusions [20th Century Democide]


2. The New Concept of Democide [Definition of Democide]
3. Over 133,147,000 Murdered: Pre-Twentieth Century Democide


4. 61,911,000 Murdered: The Soviet Gulag State
5. 35,236,000 Murdered: The Communist Chinese Ant Hill
6. 20,946,000 Murdered: The Nazi Genocide State
7. 10,214,000 Murdered: The Depraved Nationalist Regime


8. 5,964,000 Murdered: Japan’s Savage Military
9. 2,035,000 Murdered: The Khmer Rouge Hell State
10. 1,883,000 Murdered: Turkey’s Genocidal Purges
11. 1,670,000 Murdered: The Vietnamese War State
12. 1,585,000 Murdered: Poland’s Ethnic Cleansing
13. 1,503,000 Murdered: The Pakistani Cutthroat State
14. 1,072,000 Murdered: Tito’s Slaughterhouse


15. 1,663,000 Murdered? Orwellian North Korea
16. 1,417,000 Murdered? Barbarous Mexico
17. 1,066,000 Murdered? Feudal Russia
References Index

IMPORTANT NOTE: Among all the democide estimates appearing in this book, some have been revised upward. I have changed that for Mao’s famine, 1958-1962, from zero to 38,000,000. And thus I have had to change the overall democide for the PRC (1928-1987) from 38,702,000 to 76,702,000. Details here.
I have changed my estimate for colonial democide from 870,000 to an additional 50,000,000. Details here.

Thus, the new world total: old total 1900-1999 = 174,000,000. New World total = 174,000,000 + 38,000,000 (new for China) + 50,000,000 (new for Colonies) = 262,000,000.

Just to give perspective on this incredible murder by government, if all these bodies were laid head to toe, with the average height being 5′, then they would circle the earth ten times. Also, this democide murdered 6 times more people than died in combat in all the foreign and internal wars of the century. Finally, given popular estimates of the dead in a major nuclear war, this total democide is as though such a war did occur, but with its dead spread over a century.


Figure 1.1. Megamurderers and Their Annual Rates
Figure 1.2. Democide Lethality
Figure 1.3. Golgotha Among the Largest States
Figure 1.4. Golgotha’s Racial/Ethnic Composition
Figure 1.5. Regional Origin of Golgothians
Figure 1.6. Democide Compared to War Battle-Dead
Figure 1.7a. Power Curve of Total Democide
Figure 1.7b. Power Curve of War Battle-Dead
Figure 1.7c. Power Curve of Democide Intensity
Figure 1.7d. Power Curve of War Intensity (Killed)
Figure 1.8. Democide Versus War Battle-Dead; Democracies Versus Nondemocracies
Figure 1.9. Range of Democide Estimates for Regimes
Figure 4.1 Soviet Democide Components and War/Rebellion Killed 1917-1987
Figure 4.2. Soviet Democide and Annual Rate by Period.
Figure 5.1. PRC Democide and Annual Rate by Period
Figure 5.2. PRC Democide by Source
Figure 5.3. PRC Democide, Famine, and War/Revolution Deaths by Period
Figure 6.1. Nazi Democide Compared to That of Others
Figure 7.1. Nationalist Versus Communist Democide
Figure 8.1. Components of Japanese Democide in World War II
Figure 9.1. Estimated Cambodian Population Versus Predicted
Figure 9.2.. Estimated Regime Effects on the Cambodian Population
Figure 9.3. Sources of Unnatural Cambodian Deaths
Figure 9.4. Perpetrators of Cambodian Democide
Figure 9.5. Cambodian Democide Rates Compared to Others
Figure 10.1. Deaths From Turkey’s Genocide, War, and Famine 1900-1923
Figure 11.1. Comparison of Vietnam War and Post-War Deaths in Vietnam, Laos, and Cambodia 1954-1987
Figure 14.1. Democide Annual Rates: Yugoslavia Compared to Others


Table 1.1. Democratic Versus Nondemocratic Wars 1816-1991
Table 1.2. 20th Century Democide
Table 1.3. Fifteen Most Lethal Regimes
Table 1.4. This Century’s Bloodiest Dictators
Table 1.5. Some Major Episodes and Cases of Democide
Table 1.6. Democide and Power
Table 2.1. Sources of Mass Death
Table 3.1. Selected pre-20th Century Democide and Totals
Table 4.1. Overview of Soviet Democide
Table 5.1.. PRC Democide 1949-1987
Table 6.1. Selected Nazi Democide and European War Dead
Table 6.2. Nazi Democide Rates
Table 6.3.. Comparison of Nazi Democide to That of Other Regimes
Table 7.1. China’s Democide, Famine, War and Rebellion Dead, 1928-1949
Table 7.2. Period and Annual Democide Rates %
Table 8.1. Japanese Democide in WWII
Table 9.1. Cambodian Dead 1967-1978
Table 9.2. Conditions of Life Under the Khmer Rouge
Table 9.3. Cambodian Democide Rates Compared to Others
Table 10.1. Turkey’s Dead 1900-1923
Table 10.2. Turkey’s Armenian and Greek Genocide
Table 11.1. Vietnam’s War-Dead and Democide 1945-1987
Table 11.2. Vietnam War and Post-War Dead 1954-1987
Table 11.3. Vietnam: Comparative Democide Rates
Table 12.1. German Expulsion Democide
Table 12.2. German Expulsion Democide Rates
Table 13.1. Pakistan Dead March to December 1971
Table 14.1. Democide in Yugoslavia
Table 14.2. Comparison of Yugoslavian Democide and Democide Rates
Table 15.1. North Korean Democide 1948-1987
Table 16.1. Mexican Democide 1900-1920
Table 17.1. Russian Democide 1900-1917


This is my fourth book in a series on genocide and government mass murder, what I call democide. The previous works concentrated on the four regimes that have committed the most democide, specifically the Soviet Union, Nationalist China under Chiang Kai-shek, communist China, and Nazi Germany.1 This study includes the core results of those works in addition to all other cases of democide in this century up to 1987.2
Given the extent and detail of these books, the reader may be surprised that the primary purpose was not to describe democide itself, but to determine its nature and amount in order to test the theory that democracies are inherently nonviolent. They should have no wars between them, the least foreign violence and government related or directed domestic violence (revolutions, coups, guerrilla war, and the like), and relatively little domestic democide. I have substantiated the war, foreign, and domestic violence parts of this theory in previous works3 and took up the research associated with this book and its three predecessors in order to test the democide component. As will be seen, the results here clearly and decisively show that democracies commit less democide than other regimes. These results also well illustrate the principle underlying all my findings on war, collective violence, and democide, which is that the less freedom people have the more violence, the more freedom the less violence. I put this here as the Power Principle: power kills, absolute power kills absolutely.

In developing the statistics for this and the previous three volumes, almost 8,200 estimates of war, domestic violence, genocide, mass murder, and other relevant data, were recorded from over a thousand sources. I then did over 4,200 consolidations and calculations on these estimates and organized everything into tables of estimates, calculations, and sources totaling more than 18,100 rows. My intent is to be as explicit and public as possible so that others can evaluate, correct, and build on this work. I give the appendices for the Soviet, Chinese, and Nazi democide in my books on them. The appendices for this book were too massive to include here (one appendix table alone amounts to over 50 pages) and are given in a supplementary volume titled Statistics of Democide. I also include therein the details and results of various kinds of multivariate analysis of this democide and related data.

Then what is covered here? This book presents the primary results, tables, and figures, and most important, an historical sketch of the major cases of democide–those in which 1,000,000 or more people were killed by a regime. The first chapter is the summary and conclusion of this work on democide, and underlines the roles of democracy and power. Following this, chapter 2 in Part 1 introduces the new concept of democide. It defines and elaborates it, shows that democide subsumes genocidal killing, as well as the concepts of politicide and mass murder, and then tries to anticipate questions that the concept may arouse. It argues that democide is for the killing by government definitionally similar to the domestic crime of murder by individuals, and that murderer is an appropriate label for those regimes that commit democide. Readers that are satisfied with the thumbnail definition of democide as murder by government, including genocidal killing,4 can ignore this chapter. It is essential, however, for those with a professional interest in the results or wish to question the conclusions.

Following this chapter is a rough sketch of democide before the 20th century. Although hardly any historical accounting has been done for genocide and mass murder, as for the Amerindians slaughtered by European colonists or Europeans massacred during the Thirty Years War, a number of specific democidal events and episodes can be described with some historical accuracy and a description of these provides perspective on 20th century democide. I have in mind particularly the human devastation wrought by the Mongols, the journey of death by slaves from capture through transportation to the Old and New Worlds, the incredible bloodletting of the Taiping Rebellion, and the infamous Paris executions and relatively unknown genocide of the French Revolution. The upshot of this chapter is to show that democide has been very much a part of human history and that in some cases, even without the benefit of modern killing technology and implementing bureaucracy, people were beheaded, stabbed, or sliced to death by the hundreds of thousands within a short duration. In some cities captured by the Mongols, for example, they allegedly massacred over 1,000,000 men, women, and children.

Parts 2 to 4 present all the regimes murdering 1,000,000 or more people in this century, a chapter on each. These are written so as to show which regime committed what democide, how and why. The emphasis is on the connection between a regime, its intentions, and its democide. Although each of the case studies drives toward some final accounting of the democide, the specifics of such figures and the nature and problems in the statistics are ignored. These are rather dealt with in each appendix to a case study (given in Statistics of Democide), where each table of estimates, sources, and calculations is preceded by a detailed discussion of the estimates and the manner in which the totals were determined. The historical description of a case given here is only meant to provide an understanding of the democide. For this reason many specific examples will be given of the kind and nature of a regime’s killing. I have generally avoided, however, tales of brutal torture and savage killing unless such were useful to illustrate an aspect of the democide.

These chapters are ordered from the greatest of these killers to the lesser ones, as one can see from the table of contents. Part 2 presents the four deka-megamurderers, beginning with a chapter on the Soviet Union’s near 61,000,000 murdered, then including chapters on Communist China and Nazi Germany, and ending with a chapter on the now virtually unremembered killing of the Chinese Nationalist regime. Since these four regimes were the subjects of the previous three volumes,5 the four chapters simply summarize the democide and conclusions. I hope I will be excused for using Greek prefixes for labeling these regimes (deka- means ten or tens; mega- means million), but we need concepts for the various levels of government murder and there is no comparable English term (“murderer of tens of millions” is clumsy).

Part 3 presents in order the lesser-megamurders, those that have killed 1,000,000 to less than 10,000,000 citizens and foreigners. A chapter also is devoted to each. In some cases, as for Poland’s murder of ethnic Germans and Reichdeutsch, a whole series of events spanning several countries was covered. In this case Poland’s treatment of these Germans was part of a pattern of expulsion from Eastern Europe after World War II. In some cases also, several successive regimes for the same country had committed democide and these were therefore treated together, as for the Sihanouk, Lon Nol, Pol Pot, and Samrim regimes of Cambodia.

There were three regimes–those of the Czar in Russia, North Korea’s, and Mexico’s from 1900 to 1920–for which the estimates were not sufficient in number or quality to make a final determination of their democide. What estimates there were total over 1,000,000 murdered, but I treat this total as only an indictment for murder. These three are described in Part 4 as suspected megamurders.

In summary chapter 1 and in each of the case studies I present democide totals of one sort or another. With the exception of those that are directly cited from other works, how have I determined these figures, such as that Khmer Rouge regime likely murdered 2,000,000 Cambodians? The prior question is: how should these democide figures I give, totals or otherwise, by looked at? As, with little doubt, wrong! I would be amazed if future archival, historical research, and confessions of the perpetrators came up with this figure or one within 10 percent of it. Regimes and their agents often do not record all their murders and what they do record will be secret. Even, however, when such archives are available, such as after defeat in war, and they are kept by the most technologically advanced of regimes with a cultural propensity for record keeping and obedience to authority, and a bureaucratic apparatus doing the murders systematically, the total number of victims cannot be agreed upon. Consider that even after all the effort over forty-five years by the best scholars of the Holocaust to count how many Jews were killed by the Nazis, even with total access to surviving documents in the Nazi archives and the first hand reports of survivors and participants, the difference between the lowest and highest of the best estimates is still 41 percent.6

All the totals and figures in this book should therefore be viewed as rough approximations, as suggestive of an order of magnitude. This gross uncertainty then creates a rhetorical problem. How does one assert consistently and throughout a book such as this that each democide figure, as of the Khmer Rough having killed 2,000,000 Cambodians, is really a numerical haze–that we do not know the true total and that it may be instead 600,000 or even 3,000,000 that they killed? Except in cases where it is difficult to assert without qualification a specific figure (as in the chapter titles), or space and form do not allow a constant repetition of ranges, as in the summary chapter, I will give the probable range of democide and then assert a “most likely” (or “probable” or “conservative”) mid-estimate. Thus, I will conclude in chapter 9 that the Khmer Rouge likely killed from 600,000 to 3,000,000 of their people, probably 2,000,000 (this mid-value is simply a subjective probability and will be discussed shortly). All the appendices will develop and discuss such a range. For sub-totals in the historical description of a case I usually simply mention the mid-value, qualified as mentioned.

The how and why of an alleged democide range then is critical and it is not determined casually. Now, I have elsewhere published the methods that I use7 to assess the democide of a regime, and should point out here summarily that this is an attempt to bracket the unknown and precisely unknowable democide by seeking a variety of published estimates, and most important, the highest and lowest ones from pro and anti-government sources.8 I then consolidated these for different aspects of a regime’s democide, such as for summary executions, prison deaths, or disappearances, into low to high ranges. To get an overall range for a regime, as of that for the Khmer Rouge, I then sum all the consolidated lows to get an overall low democide, the consolidated highs to get an overall high.

The value of this approach lies in the great improbability that the sum of all the lowest estimates for a regime would be above the true total; or that the sum of all the highs would be below it. The fundamental methodological hypothesis here is then that the low and high sums (or the lowest low and highest high where such sums cannot be calculated) bracket the actual democide. This of course may be wrong for some events (like a massacre), an episode (like land reform), or an institution (like re-education camps), but across the years and the many different kinds of democide committed by a regime, the actual democide should be bracketed.

Within this range of possible democide, I always seek a mid-range prudent or conservative estimate. This is based on my reading of the events involved, the nature of the different estimates, and the estimates of professionals who have long studied the country or government involved. I have sought in each case the best works in English on the relevant events so that I would not only have their estimates along with the others, but that their work would guide my choice of a prudent overall estimate. The details of this effort for each case is given in the relevant appendix in the related volume, Statistics of Democide.

Given my admission that I can only come within some range of an actual democide, a range that may vary from low to high by thousands of percent, why then will I so precisely specify a democide? For example, in the chapter for communist China I will give the range of its democide as 5,999,000 to 102, 671,000, most likely 35, 236,000 people killed. Why such apparent and misleading accuracy? Why not simply make the range 5,000,000 to 105,000,000, with a mid-value of 35,000,000? This I would like to do (and have been urged by colleagues to do), but for many cases the democide figures result from calculations on or consolidations of a variety of estimates for different kinds of democide (such as for “land reform,” labor camps, and the “Cultural Revolution”). When all calculations or consolidations are added together the sum comes out with such apparent precision. That is, the low and high and 35,236,000 mid-democide for communist China’s democide are sums. To then give other than these sums can create confusion between the discussion of the cases and the appendices in which the estimates and calculations are given in detail.

I handle this presentation problem in this way. Where specification of the final democide figures calculated in an appendix is necessary, as in a table, I give them with all their seeming exactitude. Where, however, such is unnecessary, I will then round off to the first or second digit and use some adjective such as “near” or “around” or “about.” Thus, communist China’s democide was about 35,000,000.

After eight-years and almost daily reading and recording of men, women, and children by the tens of millions being tortured or beaten to death, hung, shot, and buried alive, burned or starved to death, stabbed or chopped into pieces, and murdered in all the other ways creative and imaginative human beings can devise, I have never been so happy to conclude a project. I have not found it easy to read time and time again about the horrors innocent people have been forced to suffer. What has kept me at this was the belief, as preliminary research seemed to suggest, that there was a positive solution to all this killing and a clear course of political action and policy to end it. And the results verify this. The problem is Power. The solution is democracy. The course of action is to foster freedom.


*This is a pre-publisher edited version of the “Preface” in R.J. Rummel’s Death By Government, New Brunswick, NJ: Transaction Publishers, 1994

1. Rummel (1990, 1991, 1992).

2. I started this research in 1986 and the cutoff year for the collection of data was made 1987. For consistency in comparing different cases and to avoid constantly having to change total figures as new democides occurred, I have stuck to the 1987 cutoff. This means that post-1987 democides by Iraq, Iran, Burundi, Serbia and Bosnian Serbs, Bosnia, Croatia, Sudan, Somalia, the Khmer Rouge guerrillas, Armenia, Azerbaijan, and others have not been included.

I start the 20th century with year 1900. I realize that by our calendar the 20th century really begins with year 1901. However, I was uncomfortable with including 1900 in the previous century.

3. See Rummel (Understanding Conflict and War, 1975-81; “Libertarianism and International Violence,” 1983; “Libertarianism, Violence Within States, and the Polarity Principle,” 1984; “Libertarian Propositions on Violence Within and Between Nations: A Test Against Published Research Results,” 1985). While that democracies don’t make war on each other has been verified by others and well excepted by students of international relations, that democracies have the least foreign violence has been controversial and a number of studies allege they find no difference between regimes on this. But this has been due to different and in my view inappropriate methods. I argue that the more democratic (libertarian) a regime, the more the inhibition to war or foreign violence. This therefore should be tested in terms of war’s severity-by the number of people killed either in total or as a proportion of the population. However, other’s have tested this by correlating type of regime with the number of wars it has fought. One should not be surprised, therefore, that they find hardly any correlation between regime and war, since they are treating all wars as alike, where even the tiny democratic wars such as the American invasion of Grenada and Panama or the British Falkland Islands War are given the same weight as World War I or II for Germany or the Soviet Union. In any case, one of the side results of this study is to further substantiate that democracies have the least foreign violence, i.e., that even in war democracies suffer far fewer deaths than other regimes (see Table 1.6 and Figures 1.6, 1.7b, 1.7d, and 1.8).

4. By the Genocide Convention, genocide can refer to other than killing, such as trying to destroy a group in whole or in part by taking away its children.

5. See Note 1.

6. Rummel (1992, p. 5).

7. See Rummel (1990, Appendix A; 1991, pp 309-316).

8. This has caused some misunderstanding among readers. That I use biased or ideological sources, as of communist publications on American atrocities in Vietnam or official Iraq statistics for the death toll among Kurds during the civil war, is part of my attempt to get at the lowest or highest democide or war-dead estimates. There are therefore many items in my references that no self-respecting scholar would list normally. I include them because I use their estimates and not because I believe them objective or of high quality. Moreover, the omission of a particular work from the references does not mean that I have not used it. I have consulted, read, or studied for this work many times more publications than the references list here. I have only included those I have cited in writing a chapter or those from which I have taken the estimates listed in the appendix tables. Those references listed in the Soviet, China, and Nazi democide books are not repeated here unless they also have been cited in this book.

For citations see the Death By Government REFERENCES

Many colleagues, students, and readers of previous drafts contributed to this effort through their ideas, comments and suggestions, recommendation of sources, estimates, or material they passed on to me. In particular I want to thank Rouben Adalian, Belinda Aquino, Dean Babst, Yehuda Bauer, Douglas Bond, Israel Charny, William Eckhardt, Wayne Elliott, Helen Fein, Irving Louis Horowitz, Hua Shiping, B. R. Immerzeel, Benedict Kerkvliet, Milton Leitenberg, Guenter Lewy, Heath Lowry, John Norton Moore, J. C. Ramaer, Rhee Sang-Woo, Max Singer, Spencer Weart, Christine White, and J. A. Willinge. I am especially indebted to my colleagues Manfred Henningsen and George Kent for their help and support throughout this work. I hasten to add that I alone am responsible for any errors or misconceptions that appear here.
I also am indebted to the United States Institute of Peace for a grant to my project on comparative genocide, of which this book is a part. The views expressed here are those of the author and do not necessarily reflect the views of the Institute or its officers.

Finally and not least, my ability to complete this work and the form it took owes much to my wife Grace, much more than she knows. Thanks sweetheart.



Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s