Conflict · Tigray confrontation
Ethiopian drones reportedly hit Eritrean troops in Tigray, turning former allies against each other
Witnesses and a conflict monitor describe attacks near Adigrat after a diplomatic rupture. Neither government has confirmed the strikes, and casualties remain unknown—not zero.
Transcript · LOADING
Story trackerDay 7
1 report50 outlets74 sources
The armies that fought together in Tigray’s last war are now reported to be firing on one another. Ethiopian drones reportedly struck Eritrean troops, military vehicles and ammunition around Adigrat overnight October 7–8, 2026—one week after diplomatic relations collapsed. Yet neither government had confirmed the attacks in the October 8 reporting, and the human toll remains unknown. 1 6
The distinction is important: there is evidence of a reported military confrontation, not a verified accounting of a battle. AFP’s account rests on the conflict monitor ACLED and anonymous witnesses. Their descriptions suggest repeated attacks and some destroyed trucks, but establish neither how many people were hurt nor how much equipment was lost. 1 13
The diplomatic break is firmer ground. On October 1, Ethiopia ordered its embassy in Asmara closed and declared 10 Eritrean diplomats persona non grata, citing hostility and a security threat. Eritrea severed diplomatic ties the same day in response. Those decisions explain the deteriorating relationship; they do not prove either country’s account of the subsequent fighting. 71 58
What happened along the road
Adigrat lies about 25 kilometres south of the Eritrean border in eastern Tigray. Two anonymous residents told Reuters that two military trucks carrying dozens of soldiers in off-white camouflage passed through the town on the morning of October 7. That is an observation of vehicles and soldiers—not an exact troop strength, and certainly not a count of trucks later destroyed. 12 4
ACLED reported that Eritrean forces crossed at Zalambessa, also spelled Zelambesa, and moved toward the Adigrat–Idaga Hamus road. Its October 7 alert said Ethiopian National Defence Forces drones targeted them at Guada while they travelled by vehicle toward Idaga Hamus. The National also quoted ACLED analyst Jalale Getachew Birru on that route and strike location. 36 4
There is a timing problem that prevents a clean battle chronology. ACLED published that alert at 16:30:55 UTC on October 7, but supplied no attack hour or strike count. A separate October 8 alert described forces that had crossed the previous day reportedly spending the night withdrawing from Adigrat. During that movement, vehicles carrying troops and equipment were reportedly targeted by drones believed to have been launched by Ethiopian federal forces, with heavy explosions heard into the morning. 36 50
Those are two dated accounts, not proof of two separate engagements. Guada may have been an earlier October 7 attack or part of the overnight sequence; the available reporting does not settle the overlap. Nor does the spelling “Gadua” in another account establish a second location. Adding the reports together would manufacture a strike total the evidence does not provide. 50 10
AFP reported multiple overnight strikes on Eritrean troops and ammunition around Adigrat. One witness said some Eritrean lorries had been destroyed and were no longer on the streets. Other descriptions included repeated strikes, a large explosion and troops appearing to retreat north. These accounts are consequential, but an apparent retreat does not establish that withdrawal was completed. 1 13
Known
Unknown
- No established death or injury count, verified vehicle-loss inventory or quantified ammunition loss is available.
- Civilian exposure and damage to homes or services remain unmeasured; no specific civilian casualty has been established.
Next
- Whether either government addresses the specific attacks, whether the confrontation continues, and what happened to people in and around the targeted vehicles.
Unknown casualties mean unknown casualties—not zero. The reporting also does not establish an evacuation order, road closure, aid interruption or new displacement total. For people living along this route, the absence of those measurements is a gap in the account, not reassurance that the consequences were slight. 1 4
Withdrawal demands are not strike confirmations
Ethiopia’s public demand is explicit. Enatalem Melese, minister of the Government Communication Service, told Bloomberg on October 7 that foreign military forces must leave Ethiopian territory immediately. 32 44
All foreign military forces operating within our sovereign territory must withdraw immediately. In particular, Eritrean forces operating in Tigray must leave Ethiopian territory without delay.
That statement asserts an Eritrean military presence and demands its end. It does not acknowledge an Ethiopian drone operation, establish who initiated the reported confrontation or demonstrate that anyone has withdrawn.
Eritrea has publicly rejected allegations of an Eritrean military presence in Ethiopia. Information Minister Yemane G. Meskel’s specific October 7 statement instead accused Ethiopia’s ruling Prosperity Party of constructing a justification for aggression: “The PP's overarching objective is to create a fallacious narrative of "self-defense" to rationalize and camouflage its war of aggression.” 7 23
His accusation describes an Ethiopian offensive aimed at sovereign access to the sea and a presence on Eritrea’s southern coast, including Assab. It invokes remarks attributed to an unnamed Ethiopian general, including a September 27 interview. Neither the general’s identity nor those remarks has been independently verified; they are part of the minister’s allegation, not established Ethiopian war aims. 23 38
The October 7 statement does not directly confirm or deny soldiers crossing at Zalambessa or passing through Adigrat, and it does not establish what happened during the subsequent overnight attacks. Failure to deny that particular crossing is not an admission. Neither side’s argument supplies the missing evidence about the confrontation itself. 23 12
The historical label needs equal care. These cannot be Ethiopia and Eritrea’s first-ever direct clashes: their 1998–2000 border war, associated with Badme, killed tens of thousands. The available record also does not establish that these were their first direct clashes since the 2018 rapprochement. The supported description is narrower: reported Ethiopian strikes on Eritrean military units in the current confrontation. 54 6
That rapprochement involved two agreements. Abiy Ahmed and Isaias Afwerki signed the July 9, 2018 Joint Declaration of Peace and Friendship in Asmara, declaring the state of war ended and committing to cooperation, restored connections and diplomatic relations, and implementation of the boundary commission decision. Their September 16 Jeddah agreement reaffirmed those commitments and added security, defence, trade and investment cooperation, joint economic zones and a high-level joint committee. 63 60
Eritrean forces subsequently fought alongside Ethiopian federal troops against the Tigray People’s Liberation Front during the November 2020–2022 Tigray war. But the November 2, 2022 Pretoria cessation-of-hostilities agreement was between Ethiopia’s federal government and the TPLF. Eritrea neither signed it nor was named in its text. The November 12 commanders’ implementation declaration likewise involved the federal government and TPLF, not Eritrea. Peace on that track was not a new bilateral settlement binding both states. 72 65 67 68
Claims of fighter-jet involvement, additional strike locations or a completed flight of Eritrean troops remain unverified. So does a purported Egypt–Eritrea mutual-defence obligation tied to this incident. The National reported Egyptian television host Ahmed Moussa’s claim of a joint defence agreement, but noted that no text had been made public. Political and security coordination alone cannot establish an obligation for Egypt to intervene. 4
As of October 8, the immediate question was not whether a dramatic historical label could be attached to the reports. It was whether this reported confrontation would continue—and whether an accounting of its human cost would emerge. Ethiopia had demanded immediate departure, but no separately dated withdrawal deadline or scheduled diplomatic next step had been announced. 32 4
Sources and additional supplied material
Incident reporting: Ahram Online’s AFP account and alternate article address. 1 2 Natna News. 3 The National. 4 NAMPA’s AFP report. 5 Al Arabiya’s AFP report. 6 Al Jazeera. 7 Kulu Media’s targeting and withdrawal reports. 8 11 BBC News. 9 IDN-InDepthNews’s EEPA account. 10
Further reporting: Reuters. 12 The Africa Report. 13 SEE News. 14 Daily Sabah. 15 Martin Plaut’s ACLED assessment. 16 Brussels Reporter. 17 Sight Magazine’s Reuters account. 37 Notizie Geopolitiche. 35
Statements and monitoring: Bloomberg. 32 ACLED’s October 7 post and Ethiopia and Eritrea pages. 36 50 52 Government Communication Service leadership and ministerial material. 44 45 DNE Africa’s account of Eritrea’s accusation. 23 Yemane G. Meskel’s account and October 4 statement. 38 39
Additional supplied posts, reposts and commentary, not independent confirmation of the attacks: MOI Eritrea on Facebook. 24 27 Yehdavid. 25 Amro Abdo. 26 LastNigus. 33 Alem Habteg. 40 Eri1962S. 46 Chakham Analyst. 51
Earlier withdrawal reporting and fact-checking: ReliefWeb. 18 Hiiraan Online. 19 allAfrica. 20 AFP Fact Check. 21 The Reporter Ethiopia. 22
Regional background, not proof of a mutual-defence obligation: MP-IDSA. 28 Middle East Forum. 29 The National’s summit report. 30 GlobalMilitary’s comparison. 31 Martin Plaut’s tripartite communiqué. 41 Radio Tamazuj. 42 Eritrea Focus’s maritime-pact account. 43 Additional Greek-language reporting from ΠΤΗΣΗ. 34
Additional legal and technical background, not incident verification: Oxford Public International Law. 47 Cambridge’s self-defence chapter. 48 Small Arms Defense Journal. 49
Historical and diplomatic reporting: BBC’s alternate address. 53 RFI’s AFP report. 54 Eritrea Focus’s troop account. 55 Al Jazeera’s rupture analysis and severance report. 56 58 Sean D. Murphy’s historical paper listing on SSRN. 57 WSLS’s troop-movement report. 59
Peace instruments and context: PA-X. 60 Al Jazeera’s 2018 commentary and archived news report. 61 62 Eritrean Center for Strategic Studies. 63 Africanews. 64 Reuters on Pretoria. 65 Heinrich Böll Stiftung. 66 African Transitional Justice Hub. 67 Wikipedia’s agreement entry. 68 International Crisis Group. 69
As aired 28 lines
- Ethiopia demands Eritrean forces leave its territory immediately.
- Ethiopian drones reportedly struck Eritrean troops, vehicles and ammunition in Tigray, confronting armies that previously fought together. Neither government has confirmed the attacks, and the human toll is unknown.
- What is established about this reported confrontation, and what remains unresolved about the attacks and withdrawal?
- Adigrat lies about 25 kilometres south of the Eritrean border. The reported troop movement crossed at Zalambessa, also spelled Zelambesa, toward the Adigrat–Idaga Hamus road. Guada is the reported strike location along that movement, but its precise position is not established.
- Multiple drone strikes were reported around Adigrat overnight October 7–8, 2026. Troops, military vehicles and ammunition were described as targets. The attacks have not been independently verified, and neither government had answered questions about the fighting in the October 8 reporting.
- Two military trucks carrying dozens of soldiers in off-white camouflage were seen passing through Adigrat on the morning of October 7. That observation establishes neither an exact troop strength nor how many vehicles were subsequently targeted, hit or destroyed.
- The October 7 Guada account gives no attack hour or strike count. The October 8 account describes vehicles carrying troops and equipment targeted during an overnight withdrawal, with explosions into the morning. Their overlap remains unresolved; they cannot establish a combined attack total.
- The earlier alert was published at 16:30:55 UTC, not necessarily when an attack happened. Another account spells Guada as Gadua; that variant does not establish a separately located incident. Nor does the earlier publication establish a distinct daytime engagement.
- Some Eritrean lorries were described as destroyed and no longer on the streets, with troops apparently retreating north. No verified vehicle-loss inventory, ammunition-loss quantity or death and injury count exists. Unknown casualties do not mean nobody was killed or wounded.
- Descriptions of repeated strikes, a large explosion and vehicles hit by a drone remain unverified accounts, not an independent damage assessment. An apparent retreat is not proof that withdrawal was completed. No specific civilian casualty has been established, and damage to homes or services remains unmeasured.
- Ethiopia's October 7 demand explicitly singles out Eritrean forces in Tigray for immediate departure. It asserts a foreign military presence on Ethiopian territory, but it does not acknowledge that Ethiopia launched the reported drone strikes or establish whether anyone withdrew.
- Eritrea has publicly rejected allegations of an Eritrean military presence in Ethiopia. Its specific October 7 statement, however, attacks Ethiopia's premise and justification rather than establishing what happened at the crossing or during the subsequent overnight attacks.
- Eritrea accuses Ethiopia's ruling Prosperity Party of manufacturing a self-defence rationale for aggression. The accusation alleges an offensive aimed at sovereign access to the sea and an Ethiopian presence on Eritrea's southern coast, including Assab. Those intentions are not independently established.
- The alleged Red Sea objective rests partly on remarks attributed to an unnamed Ethiopian general, including a September 27 interview. Neither the speaker's identity nor those remarks has been independently verified. Their inclusion in a ministerial accusation does not verify them.
- The October 7 statement does not directly confirm or deny soldiers crossing at Zalambessa or passing through Adigrat. Silence on that specific crossing is not an admission. Neither government's argument establishes who initiated the reported confrontation or what legal justification applies.
- Ethiopia ordered its embassy in Asmara closed on October 1 and declared 10 Eritrean diplomats unwelcome, citing hostility and a security threat. Eritrea severed all diplomatic ties the same day in response. That rupture does not prove either account of the fighting.
- Ethiopia and Eritrea fought a conventional border war in 1998–2000, associated with Badme, killing tens of thousands. Calling these reported strikes their first-ever direct clash is wrong. The available record also does not establish a first direct clash since the 2018 rapprochement.
- Abiy Ahmed and Isaias Afwerki signed the Joint Declaration of Peace and Friendship in Asmara on July 9, 2018. It declared the state of war ended, promised cooperation, restored connections and diplomatic relations, and committed to implementing the boundary commission decision.
- The September 16 Jeddah agreement reaffirmed that declaration and added security, defence, trade and investment cooperation, joint economic zones and a high-level joint committee. Both instruments were between Ethiopia and Eritrea. The later settlement ending fighting with Tigrayan forces involved different parties.
- Eritrean forces fought alongside Ethiopian federal troops against the Tigray People's Liberation Front from November 2020 through the 2022 ceasefire. That was a battlefield alignment, not an established published mutual-defence treaty. The Pretoria settlement was between the federal government and the TPLF.
- The Pretoria cessation-of-hostilities agreement was signed November 2, 2022, with a reported effective date of November 3. Redwan Hussien and Getachew Reda were lead negotiators. Eritrea was neither a signatory nor named in the text; other non-federal Ethiopian forces allied with the federal army were also unnamed.
- The November 12 follow-on implementation declaration again involved federal and TPLF commanders, not Eritrea. Peace between those parties did not create a new Ethiopia–Eritrea settlement binding both states. The supported current change is a reported direct strike on Eritrean units after diplomatic relations collapsed.
- No public treaty text establishes an Egyptian obligation to intervene in this incident. A claim that Egypt and Eritrea have a joint defence agreement remains unsupported by a published text. Political and security coordination is not, by itself, a mutual-defence commitment.
- Existing coordination includes an October 2024 communiqué with Somalia, a May 2026 maritime-transport pact, and an October 4 summit with Sudan and Somalia on Red Sea security. Those arrangements do not establish a treaty obligation bearing on the reported Adigrat and Guada attacks.
- Fighter-jet involvement and attacks at additional locations remain unverified. Claims of troops fleeing go beyond the limited description of an apparent northward retreat. Casualty figures associated with other places cannot be imported into this incident, and none confirms the reported drone attacks.
- The answer is limited but consequential: military movement and drone attacks are reported along a border-region road, but the attacks lack official confirmation. The number of engagements, damage, human toll and completion of any withdrawal remain unresolved, as does whether the confrontation continued.
- People in eastern Tigray face reported military movement and air attacks near their towns. No evacuation order, road closure, aid interruption or new displacement total is established. The civilian consequences cannot be measured from the available record.
- As of October 8, neither state had acknowledged the specific attacks. Ethiopia's demand was for immediate departure, with no separately dated withdrawal deadline or scheduled diplomatic next step announced.
Sources
74 sources50 outlets4 wire copies
Source map
Follow the evidence behind this report. The map shows its connections; the links below open the sources cited in this article.
Source map loads as you read.
Revision log
- r1First published.
