Recording Myanmar's conflict event by event — who is killed, detained, and displaced, and where.
This issue uses MCARI data for Q2 2021 only. Classification follows the MCARI taxonomy.
This issue was written after the period it covers had closed. It carries neither a discussion of a difficult coding decision nor an Outlook.
July 2021
MCARI dataset, 2021 edition · MCARI_2021_v1.0.csv
Checksum in Annex A.
MCARI_Methodology_and_Codebook_v1.0.pdf · MCARI.ORG
MCARI Research Team
Any contributor may choose not to be named, and is not asked for a reason.
This report rests on the work of Myanmar's journalists. Every event in the MCARI dataset was first reported by a news organisation working in or on Myanmar, many of them under conditions that put their staff at risk.
MCARI (2021). The Myanmar Conflict Monitor: Q2 2021. Myanmar Conflict Analysis Research Institute.
Creative Commons Attribution 4.0 International (CC BY 4.0). Copy, redistribute and adapt for any purpose, including commercially, with credit to MCARI and an indication of any changes.
This applies to the report. It is not a licence over the dataset, which is distributed separately under its own terms.
A dash marks a comparison period that does not exist.
Q1 2021 covers 59 days. Changes marked * are computed on the daily average, not on the total. Townships affected is compared on the total: a count of townships does not scale with days.
Displacement figures count recorded displacement, not distinct people. One person displaced three times appears three times.
The fatality total is a minimum. Events whose sources did not say whether anyone died carry no deaths; a death reported without a number is recorded as three; deaths recorded without a side are not apportioned. None is estimated.
The conflict in Myanmar underwent a structural shift during the second quarter of 2021 (April to June) compared to the first quarter.
While the first quarter was defined by street demonstrations and lethal crackdowns, the momentum of mass protests receded sharply in Q2. In their place, armed conflict, landmine and IED explosions and targeted violence surged, marking an evolution of the conflict toward complex armed engagement.
Unlike the first quarter, which was characterized by extreme spikes in fatalities concentrated across a handful of single days, lethal violence during this period became diffuse, with deaths recorded on every one of the 91 days.
The demographic profile of fatalities also transformed. Civilian deaths fell as a proportion of recorded deaths, with military and pro-coup personnel constituting the majority of recorded fatalities.
Alongside this expanding armed conflict, local armed groups new to the record, including the People's Defense Force, appeared for the first time. Among established ethnic armed organizations, KIO/KIA was recorded in the most armed conflict events.
Detention rates declined. Separately, the record holds its first cases of sexual violence.
Important data limitations persist. In 44% of armed conflict events, the record cannot identify at least one of the participating parties. Furthermore, some military fatality figures may rest on claims by the parties to the fighting and cannot be independently confirmed.
The primary transition in the conflict during the second quarter was a movement away from protests and crackdowns toward armed conflict and targeted violence. On a daily average basis, protests and civil unrest fell by 60% from 49.9 events per day in the first quarter to 20.2 in the second quarter, while armed conflict rose more than fivefold from 3.1 to 15.8 events per day. Concurrently, landmine and IED explosions increased about twenty-two-fold from 0.25 to 5.7 events per day, and targeted violence rose about ninefold from 0.14 to 1.2 events per day.
Week by week, the shift is clear. Civilian fatalities were heavily concentrated in the first two weeks, recording 204 deaths, including 82 in a single incident in Bago on 9 April. Following that period, civilian deaths declined to between 20 and 55 per week. By contrast, military and pro-coup fatalities peaked during the final weeks of May, reaching 163 and 208 deaths per week. Protest events also diminished sharply, falling from 236 in the first week of May to 73 in the final week of June.
By month, May was the most severe of the period, recording 2,147 events and 816 fatalities (compared to 1,552 events and 490 fatalities in April, and 1,426 events and 699 fatalities in June). Compared with the first quarter, although the civilian share of fatalities declined, the total daily fatality rate rose by 63% from 13.5 to 22.0 deaths per day, with this increase driven by armed conflict events.
The second quarter recorded 1,440 armed conflict events, including 669 armed clashes, 518 landmine and IED explosions, 109 instances of artillery shelling, and 74 airstrikes. The 1,834 protest events carry no deaths: MCARI records a death at a protest as the security forces' attack on the protesters (Violence against Civilians), not as part of the protest. On a daily average, civilian deaths in violence against civilians events fell from 9.3 to 4.5 a day, while rising in targeted violence (from 0.12 to 1.1) and armed conflict (from 0.78 to 1.3).
Looking at the share of events recording a death (not Table 1's deaths per 100 events), 85.0% of targeted violence events and 34% of violence against civilians events carried a fatality, compared to 17% for armed conflict (Table 1). By contrast, only 1.0% of coercion events recorded a death, and no protest event did. An event count is not a measure of violence.
| Event type | Events | Share | QoQ | YoY | Fatalities | Share of fatalities | Fatalities per 100 events |
|---|---|---|---|---|---|---|---|
| Protests / Civil Unrest | 1,834 | 35.8% | −59.6%* | — | 0 | 0.0% | 0.0 |
| Coercion | 927 | 18.1% | +21.2%* | — | 10 | 0.5% | 1.1 |
| Violence against Civilians | 628 | 12.3% | +7.5%* | — | 423 | 21.1% | 67.4 |
| Armed Conflict | 1,440 | 28.1% | +407.1%* | — | 1,433 | 71.5% | 99.5 |
| Strategic Developments | 168 | 3.3% | +54.2%* | — | 0 | 0.0% | 0.0 |
| Targeted Violence | 113 | 2.2% | +785.7%* | — | 135 | 6.7% | 119.5 |
| Other | 15 | 0.3% | +100.0%* | — | 4 | 0.2% | — |
| Total | 5,125 | 100.0% | — | — | 2,005 | 100.0% | — |
Q1 2021 covers 59 days. Changes marked * are computed on the daily average, not on the total.
Where an event type records fewer than 20 events in the period, the rate cell reads — and the counts are still shown.
The rates in this column run from 0.0 to 119.5 per hundred events, so no total is shown: a single rate across a record in which most events record no death describes no event that occurred.
A zero means nothing was recorded, not that nothing happened.
In monthly composition, armed conflict rose from 22.9% of all events in April to 31.8% in May. As protest activity continued to diminish, armed conflict (28.2%) and protests (27.9%) stood at nearly identical shares of total events by June (Figure 1).
All seven event types are plotted, including those at or near zero. A row set in grey moves two percentage points or fewer across the period; the rest carry the slot colour. The last month's point is filled; earlier points are hollow.
The second quarter recorded 2,005 fatalities, accompanied by a marked transformation in the composition of the dead (Table 2). Military and pro-coup forces accounted for 1,137 deaths (56.7%), and non-civilians formed the majority of recorded fatalities. On a daily average basis, military fatalities increased more than fourfold, rising from 2.7 deaths per day in Q1 to 12.5 in Q2.
| Month | Military and pro-coup | Armed opposition | Civilians | Not attributed | Total | Civilian share |
|---|---|---|---|---|---|---|
| April 2021 | 206 | 11 | 273 | 0 | 490 | 55.7% |
| May 2021 | 496 | 42 | 180 | 98 | 816 | 22.1% |
| June 2021 | 435 | 35 | 191 | 38 | 699 | 27.3% |
| Period | 1,137 | 88 | 644 | 136 | 2,005 | 32.1% |
Where total fatalities in a month are below 20, the share carries no trend statement in the prose and any chart point is plotted hollow.
Conversely, 644 civilian fatalities were recorded (32.1%), representing a sharp decline from the 76.1% civilian share observed in the first quarter. Daily average civilian fatalities dropped by 31% compared to Q1. The remaining fatalities comprised 88 members of armed opposition groups and 136 individuals whose affiliation could not be determined.
The distribution of fatalities across events was more diffuse than in the first quarter, with deaths recorded on every one of the 91 days (Figure 2). While half of all fatalities in Q1 were concentrated in 33 events, reaching the same threshold in Q2 required 52 events (Figure 3). However, military and pro-coup fatalities remained heavily concentrated, with nearly half (48%) occurring in just 20 events.
The 563 events that record a death are 11.0% of all events in the period and carry all 2,005 recorded deaths. The bars above show how much of that total falls in the deadliest few.
An analysis of the five deadliest days shows that four were dominated by armed conflict, which accounted for over 90% of the deaths on those days, led by 31 May with 138 fatalities (Table 3). The sole exception was 9 April, where all 84 recorded deaths were civilians, including 82 from a single incident in Bago.
| Date (2021) | Deaths | Of them civilians | Events recorded | Townships recording a death | Armed-conflict share of that day’s deaths |
|---|---|---|---|---|---|
| 31 May | 138 | 13 | 75 | 9 | 91.3% |
| 9 April | 84 | 84 | 27 | 2 | 0.0% |
| 22 May | 78 | 6 | 68 | 9 | 92.3% |
| 5 June | 59 | 25 | 48 | 7 | 91.5% |
| 6 June | 58 | 2 | 36 | 5 | 96.6% |
Among the ten deadliest single events, nine were armed clashes or landmine incidents, with the Bago incident standing as the only instance of violence against civilians (Table 4). The highest death tolls among armed clashes, recorded in Demoso (81 deaths) and Kani (61 deaths), may rest on claims by the parties, as other military death figures may, and are given as the sources reported them.
| Date (2021) | Township | Type | Deaths | Of them civilians |
|---|---|---|---|---|
| 9 April | Bago, Bago-East | Attack on civilians | 82 | 82 |
| 31 May | Demoso, Kayah | Armed clash | 81 | 0 |
| 22 May | Kani, Sagaing | Armed clash | 61 | 0 |
| 6 June | Mindat, Chin | Armed clash | 42 | 0 |
| 14 April | Momauk, Kachin | Landmine / IED | 32 | 0 |
| 1 June | Myawaddy, Kayin | Armed clash | 32 | 0 |
| 6 May | Momauk, Kachin | Armed clash | 30 | 0 |
| 31 May | Pauk, Magway | Armed clash | 30 | 0 |
| 24 June | Katha, Sagaing | Armed clash | 30 | 0 |
| 23 May | Demoso, Kayah | Armed clash | 26 | 0 |
The geography of the conflict varied markedly across states and regions during the second quarter (Map 1). Sagaing Region recorded the highest fatality total nationwide at 531, followed by Chin State with 283, Kachin State with 226, Kayah State with 173, and Kayin State with 158.
Circle size is the number of recorded events; fill is the number of recorded fatalities. A large unfilled circle is a township with many events and no recorded death. 172 of 5,125 events could not be resolved to township level and are not plotted.
Boundaries: MIMU 250k admin-3 (township) polygon boundaries, valid on 15 February 2024. Projection: Albers equal-area conic (d3.geoConicEqualArea).
Among these areas, Chin State exhibited a distinct pattern. Despite recording 283 fatalities, only 6 were civilians, while 179 were military and pro-coup forces, 22 were armed opposition, and 76 remained unattributed. Fatalities were concentrated primarily in Mindat and Hakha townships. Furthermore, Chin State recorded 48,506 displacement events, the highest of any state or region. Conversely, in Bago-East, all 94 recorded fatalities were civilians, including 82 from the 9 April incident in Bago city. In Yangon Region, although the daily average event count fell by 42%, the division recorded 164 landmine and IED explosions, the highest volume of this event type nationwide.
A comparison of the most affected townships shows a divergence between locations with the highest event counts and those with the most severe fatalities (Table 5). In event volume, Momauk in Kachin State ranked highest with 174 events, whereas Demoso (160 deaths), Mindat (155 deaths), and Kani (142 deaths) led the fatality rankings. Across the entire period, 270 townships recorded conflict activity, with 34 townships entering the record for the first time and 32 dropping out compared to the previous quarter.
| Most events | Most fatalities | |||||||
|---|---|---|---|---|---|---|---|---|
| Township | Region or state | Events | Fatalities | Township | Region or state | Events | Fatalities | |
| Momauk | Kachin | 174 | 124 | Demoso | Kayah | 82 | 160 | |
| Hpakant | Kachin | 144 | 22 | Mindat | Chin | 51 | 155 | |
| Mogoke | Mandalay | 119 | 18 | Kani | Sagaing | 91 | 142 | |
| Kale | Sagaing | 108 | 52 | Momauk | Kachin | 174 | 124 | |
| Yinmarbin | Sagaing | 105 | 83 | Bago | Bago-East | 37 | 88 | |
| Monywa | Sagaing | 94 | 13 | Hpapun | Kayin | 59 | 85 | |
| Kani | Sagaing | 91 | 142 | Yinmarbin | Sagaing | 105 | 83 | |
| Dawei | Tanintharyi | 87 | 6 | Hakha | Chin | 33 | 72 | |
| Kutkai | Shan-North | 87 | 34 | Pauk | Magway | 13 | 63 | |
| Kyaukme | Shan-North | 87 | 3 | Myawaddy | Kayin | 32 | 60 | |
3 townships appear in both rankings.
Tamu also recorded 60 deaths.
Township-level figures use only events resolved to township level and are grouped by region and township together. 172 events in this period could not be resolved to township level and are excluded from this table.
A notable development in the second quarter was the entry of 14 organizations new to the record, including the People's Defense Force (PDF). Counting either side of an event, the People's Defense Force appeared in 49 (as the acting party in 23, Table 6), Chinland Defense Force in 21, CDF (Mindat) in 20, Karenni People's Defense Force in 19, and CDF (Hakha) in 14. Among named ethnic armed organizations, KIO/KIA was on either side of 370 armed conflict events (Annex C, Parties on armed conflict events), making it the second most frequent named participant after the military. Pro-military demonstrators (Pro-Military Protesters), which appeared on either side of 31 events in the first quarter (28 as the acting party), were entirely absent from the record during this period.
The Myanmar Military was recorded as the acting party in 1,892 events, of which 187 recorded civilian fatalities, totaling 418 civilian deaths (65%) (Table 6). In events recording an Unidentified Armed Group as the acting party, 164 civilian deaths (25%) were recorded, while 5 were recorded in events involving the Police.
| Acting party | Events | Events carrying a civilian death | Civilian deaths |
|---|---|---|---|
| Myanmar Military | 1,892 | 187 | 418 |
| Protesters | 1,711 | — | — |
| Unidentified Armed Group | 805 | 123 | 164 |
| KIO/KIA | 173 | 2 | 2 |
| Police | 167 | 5 | 5 |
| Unidentified Actor | 61 | 19 | 20 |
| SSPP/SSA-N | 36 | 1 | 1 |
| RCSS/SSA-S | 31 | — | — |
| People's Defense Force | 23 | — | — |
| Combined Group | 16 | — | — |
| (no acting party recorded) | — | 2 | 7 |
| Rioters | — | 2 | 5 |
| Yaw Defense Force | — | 1 | 5 |
| Kalay Region Defense Association | — | 1 | 3 |
Blanks are excluded: 6 events record no acting party. 112 distinct acting parties appear in the period.
This is not a ranking of responsibility. The first column counts how often a party is recorded as acting; the second and third count what the record attributes to it. They are different measures.
Examining participant profiles by conflict layer, 422 of the 433 fatalities recorded in the demonstration and crackdown layer were civilians, whereas in the armed conflict layer, 1,120 of the 1,433 fatalities were military and pro-coup forces (Table 7).
| Protest and crackdown | Armed Conflict | Outside both layers | |
|---|---|---|---|
| Events | 3,389 | 1,440 | 296 |
| Deaths | 433 | 1,433 | 139 |
| Military and pro-coup | 10 | 1,120 | 7 |
| Armed opposition | 0 | 85 | 3 |
| Civilians | 422 | 118 | 104 |
| Not attributed | 1 | 110 | 25 |
Protest and crackdown events were most concentrated in Mandalay and Sagaing; armed conflict in Kachin and Yangon.
Event types outside both layers: Other, Strategic Developments, Targeted Violence.
Important data limitations persist: 44% of armed conflict events involved an Unidentified Armed Group whose specific organizational identity cannot be determined from the record (Annex C, Parties on armed conflict events). These entities cannot be coded directly as local defense forces or PDFs, and are given as the record holds them.
The rate of detention fell sharply in the second quarter compared to the first quarter (Table 8). Daily average detentions dropped by 58%, falling from 105.6 individuals per day in the first quarter to 44.0 in the second quarter.
| Indicator | April 2021 | May 2021 | June 2021 | Period | QoQ | YoY |
|---|---|---|---|---|---|---|
| People detained | 1,311 | 1,643 | 1,054 | 4,008 | −58.3%* | — |
| Recorded displacement events | 42,563 | 95,874 | 29,581 | 168,018 | +146.9%* | — |
| Defections and surrenders | 0 | 0 | 0 | 0 | −100.0%* | — |
| People released | 95 | 169 | 130 | 394 | −98.9%* | — |
Q1 2021 covers 59 days. Changes marked * are computed on the daily average, not on the total.
Displacement figures count recorded displacement, not distinct people. One person displaced three times appears three times.
Townships recording the highest numbers of detained individuals were Myingyan (182), Mogoke (177), and Kani (121) (Figure 4). The record holds 394 individuals released from detention, representing an apparent 98.9% decline from the previous period; however, this divergence is driven by the mass amnesty on 12 February (23,314 individuals) included in the first-quarter total. Excluding the mass amnesty, 937 individuals were released in the first quarter. Against that baseline, average daily releases still fell by 73%, dropping from 15.9 to 4.3 individuals per day.
These ten townships account for 1,164 of the 4,008 people recorded detained in the period. 61 sit on events that could not be resolved to township level.
During this period, 168,018 displacement events entered the record, with over half resting on five large-scale estimates, including 35,000 in Hakha Township on 19 May and 30,000 in Hpapun Township on 6 May. Defections require explicit identification of the receiving entity to be coded, and no qualifying defections were recorded during the quarter (zero events).
Other dimensions of the conflict appeared in the data during this period: 16 territorial control changes in favour of armed opposition (including 8 in Kachin State) and 6 instances of sexual violence entered the record.
Interpretation of the findings in this quarterly assessment must account for the following data constraints.
The record names no party for over 90% of landmine and IED explosions, and in 44% of armed conflict events it cannot identify at least one participating party.
Some military and pro-coup fatality figures may rest on claims by the combatant parties and cannot be independently confirmed.
Displacement figures are counts of recorded displacement events, not of individuals, with over half of the second-quarter total dependent on estimates from five single events.
The fatality total of 2,005 is a minimum-confirmed figure.
These figures are computed from the MCARI 2021 Edition (v1.0) dataset, based on reports from national, exile, and ethnic media outlets. In 664 events the sources did not say whether anyone was killed; these carry no deaths. Recorded displacement events (168,018) track the number of times displacement was reported, not the number of distinct individuals. Comparisons with the first quarter, which covers 59 days, are made on daily averages.