Civil War in Guangxi
The Cultural Revolution on China's Southern Periphery
Andrew G. Walder



On July 16, 1968, the final battle for control of Guangxi’s capital city of Nanning was about to begin. Rebel forces had overthrown civilian governments across the province at the outset of 1967, and in the succeeding months armed factions fought one another across the region, intensifying as the long stalemate wore on. Several thousand fighters from one of two rebel alliances—the Allied Command (lianzhi)—took up positions that sealed off a downtown district around Liberation Road. The mixed commercial and residential neighborhood was located on the northern bank of a sharp bend in the Yong River, a major shipping channel. It was bordered on the west and north by a tributary known as Chaoyang Creek, which curved around the district, creating a peninsula. Dug into the neighborhood, trapped behind defensive positions, were the leaders, supporters, and surviving combat brigades of the other alliance, the April 22 faction (si.erer, or “4.22”), who had refused to surrender and were making their last stand.1

As the Allied Command forces prepared for their assault, they launched a campaign to “cleanse the population” in areas of the city under their control. The “cleansing” sought to capture members of the April faction who had fled to the city from surrounding rural districts in recent months, as their faction was suppressed. All were treated as enemy combatants. Of the 283 individuals who were arrested in the initial sweep that day, 33 were summarily executed, and the rest went missing, their fate unknown. The killings foreshadowed the brutality of the battle to come.

The Allied Command, backed by People’s Liberation Army (PLA) soldiers from the Guangxi Military District and rural militias from surrounding counties, began by hurling makeshift firebombs (glass bottles filled with gasoline) at buildings occupied by their enemies. April faction fighters responded with sniper fire and their own firebombs, and homes and shops were set ablaze. Firefighters who rushed to the scene were targeted by snipers, and three of them were killed. Allied faction forces on both sides of the Yong River bombarded boats used by the April faction to ferry supplies and reinforcements. They scored a direct hit on a freighter containing more than a thousand barrels of diesel and aviation fuel, setting the docks ablaze.

Rifle fire, firebombs, and destruction along the riverbanks failed to dislodge the defenders, so PLA troops joined the battle, bringing to bear heavy weaponry. In late July, the attacking forces used anti-aircraft machine guns, rocket-propelled grenades, and bazookas to dislodge the defenders. The strategy was highly effective when focused on individual buildings. On July 31, troops backed by an artillery unit turned their firepower onto an exhibition hall occupied by the April faction. From midafternoon into the night, they poured thousands of rounds from rifles and machine guns into the building, along with hundreds of artillery shells. The battered survivors surrendered the next morning. The assault left 23 dead and 57 wounded among the defenders; 470 were taken prisoner. The attackers suffered 6 dead and 52 wounded.

Similar weapons were then used indiscriminately to bombard the neighborhoods under April faction control. Despite the inevitable destruction, the bombardment continued, and reinforcements from the PLA and rural militias continued to join the battle. By the first days of August, shops and homes along twelve streets, covering thirty-three city blocks, were reduced to rubble, and fires blazed uncontrolled in the structures that remained standing. By August 5, after a siege of almost three weeks, April faction resistance had collapsed. The last few holdouts surrendered on August 8. The devastation to the Liberation Road district was total, and the April faction was crushed. Shops, homes, factories, government offices, and temples were destroyed. More than 10,000 families—50,000 residents—were left homeless.

The authorities would later document close to 1,600 deaths in the battle for the district and a handful of other sites in the city, a count that was admittedly incomplete. In the Liberation Road neighborhood, 680 corpses resulting from the fighting were eventually dumped into a coal pit, and a similar number were recovered from burned-out buildings. Another 72 were buried hastily along the banks of Chaoyang Creek; 52 were killed while trying to flee; and an unknown number of corpses floated off down the creek and into the Yong River. During the entire course of the battle of Nanning, only 60 Allied faction fighters were killed, along with 26 PLA soldiers.

The ordeal of the surviving defenders was just beginning. Close to 6,500 April faction fighters were taken prisoner, along with some 2,500 residents suspected of supporting them. The victors did not have a plan to process so many prisoners. The detainees were assembled under guard at several locations nearby, to be marched in groups to the government headquarters for investigation and sentencing. During one such procession Allied faction guards pulled twenty-six prisoners out of the ranks and summarily executed them.2 Large numbers of the survivors were members of the April faction from rural counties who had moved into Nanning to seek protection or to reinforce their allies in the battle for the city. A total of close to 7,000 either had surrendered after the battle or had been captured during the “cleansing” of the city’s population. They were expelled from Nanning and taken back to their home counties, which were no more forgiving: 2,324 were executed, and 246 were given long prison sentences.

The pacification of Nanning was one of two decisive urban battles that spelled the end of factional warfare in Guangxi. As the April faction was being crushed in the provincial capital, a much longer and more evenly matched battle was coming to a similar conclusion in Guilin.3 This former provincial capital in the northeastern corner of Guangxi, a region renowned for the beauty of its steep limestone mountains, had long been the stronghold of the April faction. The Guilin branch of the April alliance, known locally as the Rebel Army (zaofan dajun), had seized power in January 1967 with the backing of a PLA regiment stationed there. Relying on continuing PLA support, the Rebel Army staged a massive arms seizure from military stockpiles in August 1967. Heavily armed with military weaponry, they drove their opponents out of the city and into the surrounding counties. Not until June 1968 would the Allied Command, backed by rural militias from surrounding counties, launch their offensive to take Guilin.

Unlike the final battle for Nanning, Guilin was a prolonged, two-month campaign. Both sides had advanced weaponry—automatic rifles, light and heavy machine guns, rocket launchers and mortars. Each side had infantry and artillery companies. Each had advisors from different units of the PLA. After a cease-fire in April 1968, followed quickly by the establishment of a new city government dominated by the Allied faction, the truce broke down and the two sides fortified strongholds in different districts. Drawing on instruction from their respective military backers, the two sides organized combat units. Still smaller and less well-armed than their opponents, the Allied Command called upon rural militias in the surrounding counties, under the direction of local People’s Armed Departments (PADs), to send reinforcements. The first militia contingents entered Guilin on June 4, and battles between the two sides escalated. By the end of June, more than 8,000 militia fighters from twelve counties had entered the city. The rural militias, however, were only lightly armed. In one of the large battles that month, ninety-three were killed on the two sides, and the Rebel Army forced the militia to retreat to suburban counties.

After these initial engagements, the Allied Command realized that their weaponry was inadequate. At the end of June, they staged raids on a large PLA arms depot located at a rail terminus in Xing’an County, on the line running north from Guilin into Hunan Province. Eventually mobilizing more than 1,000 fighters, they made ten unsuccessful assaults on the depot before prevailing. They hauled off a large stockpile of arms and ammunition, including 60 cannon, more than 11,000 artillery shells, 12,000 hand grenades, and other military supplies.

The arms were distributed to militia units, and two weeks later they began a coordinated assault, advancing into Guilin from three directions. The two sides faced off on a long front, and intense fighting by infantry and artillery units continued from July 13 to 24. Enjoying a continued flow of reinforcements from rural militias, the Allied Command gradually gained the upper hand in a series of decisive battles from July 27 to August 3. They finally overwhelmed the Rebel Army forces, declaring victory on August 6.

During the battle for Guilin, the two sides mobilized more than 10,000 armed fighters. They possessed more than 13,000 rifles, 80 cannon, 19,000 cannon shells, 230,000 hand grenades, and 8.3 million rounds of ammunition. Over the most intense weeks of fighting, Guilin was paralyzed. Transportation was halted; factories, schools, and shops were closed; suburban farmers were unable to tend crops; and civilian noncombatants were terrorized. A partial count of casualties listed 406 deaths among urban combatants and noncombatants, and 198 from the militias drawn from the surrounding counties. It was common for prisoners to be summarily executed. Rural militia members who returned home after suffering casualties in Guilin exacted revenge on April faction adherents and others in order to “consolidate the rear area.” Militia forces engaged in a wave of mass killings in rural counties across Guilin Prefecture, liquidating close to 8,500 noncombatants during June, July, and August.

The Rebel Army suffered violent retribution after their defeat. The authorities initiated a coordinated sweep of neighborhoods and workplaces, focusing on Rebel Army activists and fighters, along with their many supporters. Arrests and house searches claimed more than 10,000 victims; 7,000 were held at one point in the main detention center. Prisoners were subjected routinely to beatings, torture, and public humiliation. Some were stoned to death, others beaten to death in public rallies, and there were group executions by firing squad. By the end of the suppression campaign, 345 deaths were officially recorded: 160 beaten or shot to death; 94 killed during public rallies or processions; and 91 suicides.

As the urban warfare in Nanning and Guilin reached its violent conclusion, rural counties across Guangxi were also in the midst of intensely violent upheavals, but of a very different kind. In almost all counties, factions aligned with the April coalition had been subdued months earlier by militia forces under the command of PADs. Yet a startling wave of violent persecutions intensified across Guangxi’s rural regions just as the final battles in Nanning and Guilin were reaching a climax. The casualty counts dwarfed those in the major cities, however, and the persecutions did not focus exclusively on April faction forces and their supporters. They also targeted uninvolved individuals from politically stigmatized “class enemy” households, the surviving members of former propertied classes and those historically associated with the Nationalist Party and other opponents of the new regime, along with their offspring. Moreover, the death tolls were rarely generated by armed combat or retribution by victorious militias over their vanquished enemies. Instead, unarmed village residents were liquidated in large numbers, at times in the most gruesome manner.

This was a form of violence rarely seen in urban combat and its aftermath. The victims and their killers were residents of the same rural districts. Most disturbing to those who subsequently chronicled these events was the discovery that entire families were sometimes liquidated, including the elderly and small children, in acts that were at times accompanied by sexual violence, the mutilation and display of corpses, and even cannibalism.

Emblematic of this wave of rural violence was Binyang County, northeast of Nanning, whose closest border was only 25 miles from the city’s near suburbs.4 On July 24, as the fighting in Nanning was reaching its climax, the county authorities organized meetings of public security and militia offices in neighborhoods and villages, ordering them to draw up lists of class enemies and other political suspects. Commune and village militia officers were instructed to compile “dictatorship” lists—those in villages who were to be subjected to public trials and executions. Under the direction of these committees, waves of mass killings spread across villages. Over a ten-day period from the end of July to early August, 3,691 individuals were killed or committed suicide. Of these, 2,036 were members of stigmatized “class enemy” households, few of whom had dared to get involved in factional activities. There were 191 households in which more than one individual was killed; 176 children were orphaned; and in 14 extreme cases, everyone in the family was killed. A summary paragraph in an official investigation report reflects the sense of shock felt by those who chronicled these events two decades later:

During the few days that the massacres reached a climax, there were corpses littered all over the county seat, to the point where vehicles could not pass on the streets. Families were scattered and broken up; some were dug up by the roots and completely eradicated; some were plundered by the killers so there was no home to which to return; wives were raped by their husband’s murderers or forced to submit to them.5

Detailed accounts from Binyang and other counties describe events in hundreds of individual villages that illustrate the way these campaigns were conducted. Numbing accounts of savage cruelty have been highlighted in a range of publications over the years.6 One example shows how these campaigns spread widely across rural regions. On July 29 the battalion commander of the Luojiang Commune militia returned from a meeting at the Binyang county seat and called for a meeting of the commune’s leaders. They drew up a list of eight men from several villages for execution. On the next day the commanders of the village militia escorted the men to a mass rally attended by 200 to 300 residents, during which they were harangued in speeches that detailed the individuals’ alleged crimes and called for the elimination of class enemies. At the end of the rally, members of the militia and other activists dragged the eight men to the outskirts of the village and clubbed them to death. Two days later, on August 2, the commune’s leaders met again to discuss how to push their campaign against class enemies to an even higher level. The head of the district’s recently established Revolutionary Committee gave a speech urging the commune to make more progress. He claimed that they had not done enough, and a commune of their size should “do [gao]” at least 100. The commune committee then hastily drew up a new list of six names for execution and ordered the militia commanders to bring them to the commune seat for another mass rally. At the end of the rally the individuals were dragged to the commune’s tile kiln and clubbed to death on the spot.7

As the disastrous consequences of the Cultural Revolution were chronicled and publicized by the new leaders of the Chinese Communist Party in the late 1970s and well into the 1980s, Guangxi earned a reputation as the most violent region of China. There had been urban warfare in other cities that resulted in extensive devastation and large death tolls, but destruction on the scale of Nanning and Guilin was rare.8 Violent retribution against defeated factional enemies was common elsewhere, but it rarely generated death tolls on the scale observed in Guangxi.9 There were also mass killings of “class enemy” households in other regions, but these were isolated cases, restricted to one county or a small cluster of them. In no other province were such killings as widespread and intense as in Guangxi.10 And in no other region was sexual violence and cannibalism reported with such disturbing frequency.

The most credible estimate of the total number of people killed in political violence in China during this phase of the Cultural Revolution is in the range of 1.6 million.11 A detailed series of investigations in Guangxi organized by the Beijing authorities during the 1980s documented a final count of 89,810.12 With a population of 25 million in 1967, this would yield a death rate of 3.6 per thousand in Guangxi. China’s total population at the time was close to 750 million: excluding Guangxi, the national death rate would be close to 2.1 per thousand. By this measure, Guangxi’s death rate was 70 percent higher than that of the rest of the country. There is some uncertainty, however, about how the Guangxi investigations handled the thousands of people who “disappeared.” A memoir by an official who participated in the investigations stated that there were 20,000 or so individuals recorded as “missing”—they disappeared during this violent period, but their deaths could not be documented.13 If only half of this number is added to the official death toll, it would approach 100,000, which would be a death rate of close to 4.0 per thousand, almost double that for the rest of China.

Some regions in Guangxi suffered much more intense violence. In the most violent fifth of the cities and counties, the average death rate was 7.7 per thousand. In two counties more than 10 per thousand, or 1 percent, were killed.14 The only other region of China where death rates are suspected of approaching these levels is Inner Mongolia, for which detailed evidence is scarce.15

A death rate of 3.6 to 4.0 per thousand would place Guangxi within the range of estimates for the notoriously brutal counterinsurgency campaign in Guatemala in the early 1980s or the widespread massacres of suspected communists and other leftists by the military junta in Indonesia in 1965 and 1966, both of which occurred over a much longer period of time.16 Although events in Guangxi resembled civil wars in other settings, the death rates in Guangxi are a far cry from much longer conflicts between heavily armed combatants of the kind that has accompanied civil wars in collapsing nation-states. The death rate in Bosnia-Herzegovina after the fall of Yugoslavia in the 1990s was 25.5 per thousand. The civil war that accompanied the collapse of Syria in recent years has generated death tolls of similar magnitude.17

Although Guangxi’s civil conflicts do not rank among the deadliest in recent world history, they generated a startlingly large death toll in a very short period. They did so without prolonged military campaigns between well-organized and well-equipped military units. The violence is especially puzzling, given the sudden formation of two civilian factions; their lack of an overt identification with class, ethnic, or national identities; and the fact that both sides claimed absolute loyalty to the central state as embodied in its Chairman, Mao Zedong. What made Guangxi so different from most other regions of China?


1. This account of the battle of Nanning draws on two sources: Guangxi Cultural Revolution Chronology (1990, 106–16), and Guangxi Party Committee (1987, 13:106–11).

2. Photographs of the devastated neighborhood, ranks of marching prisoners, and the summary executions were included in a publication that has long since been withdrawn in China (Guangxi Cultural Revolution Chronology 1990). Several of the photographs are reproduced in Walder (2015).

3. This account is based on Guangxi Party Committee (1987, 3:647–55, 6:242–60 and 297–306).

4. This account is based on Guangxi Party Committee (1987, 2:31–63).

5. Guangxi Party Committee (1987, 2:62).

6. For example, Sutton (1995); Zheng (1996); Su (2011); Song (2002).

7. This account is based on Guangxi Party Committee (1987, 2:40–41).

8. One well-documented example was a month-long battle in the Yangzi River port city of Luzhou, in Sichuan Province, in July 1968, which engaged some 24,000 fighters and claimed 2,000 lives (Walder 2019, 169–70; drawing on Luzhou City Annals 1998, 38; and Sichuan Province Annals 1999, 139–40).

9. Walder (2019, 178–87).

10. Two widely reported instances were in Changping and Daxing counties, in the Beijing suburbs, and in Dao County, Hunan, and several surrounding counties in Lingling Prefecture (Song 2002; Tan 2010; 2017).

11. This is based on a detailed examination of numbers reported in published and unpublished local accounts from more than 2,200 local jurisdictions, which yielded estimates that tally closely with reports of an internal Central Committee investigation conducted in the 1980s (Walder 2014; 2019, 188–90).

12. Guangxi Party Committee (1987, 7:127).

13. Yan (2012). The total death toll of 89,810 in the official report was said to include an unspecified number of “missing.” It is possible that some of the missing were included in the final official tally, which is higher than the results of tabulations mentioned in the previous paragraph. The separate reports for cities, counties, and organizations only rarely provide totals for the missing. Yan appears to refer to information about the missing that was excluded from the final report.

14. Fengshan County’s death rate was 1.36 percent; Shangsi County’s was 1.44 percent.

15. Of the 2,246 counties and cities for which officially published or unpublished data are available, 52 localities had death rates above 4.0 per thousand. Of these, 30 were in Guangxi, and the next largest number were in Inner Mongolia, where 3 counties reported death rates greater than 10 per thousand, or 1 percent (Walder 2019, 192).

16. Ball, Kobrak, and Spirer (1999, 119); Robinson (2018, 120–21).

17. For Bosnia, Zwierzchowski and Tabeau (2010). Syria’s prewar population was close to 21 million, slightly below that of Guangxi in 1966. The United Nations and other sources estimate a range of 400,000 to 500,000 dead in the Syrian upheavals.