The Soldier Who Wouldn't Leave: Eritrea's Endless Service and the Border

On a Tuesday afternoon in March 2024, Petros Haile was standing guard at a checkpoint seventeen kilometres from the Ethiopian border. He had been standing guard, in various forms, for twelve years. He was twenty-nine years old. The conscription notice that brought him to Sawa military training camp in 2012 said his service would last eighteen months. That was the law. But the law, in Eritrea, had become a formality that no one discussed.

Petros had not seen his family in Asmara in four years. His monthly salary was 500 nakfa — approximately eleven U.S. dollars at the unofficial exchange rate. He slept in a barracks with forty-three other men. The checkpoint where he worked controlled access to a road that led to nothing in particular: a few villages, some farmland, the escarpment that dropped toward the Tigray region of Ethiopia. Most days, fewer than twenty vehicles passed. On some days, none.

What made that Tuesday different was that Petros decided to leave. He walked away from the checkpoint at 14:00, crossed the border at dusk, and surrendered to Ethiopian soldiers on the other side. He knew what this meant: he could never return to Eritrea. His family would be fined. They might lose their ration cards. He did it anyway. By the time he reached Mai Aini refugee camp in Tigray three days later, he was conscript number 847 to arrive that month.

The Service That Never Ends

Eritrea's national service system was established in 1995, two years after independence. The law specified eighteen months of service: six months of military training, twelve months of deployment. The stated purpose was nation-building. Every citizen, male and female, between eighteen and forty would contribute. It was framed as temporary, necessary, patriotic.

But after the 1998-2000 border war with Ethiopia, the system changed. The government declared a state of emergency. Conscripts were not discharged. New cohorts were called up and not released. By 2002, the eighteen-month limit had become a legal fiction. By 2010, it was understood that national service meant indefinite service. No one knew when — or if — they would be released. The government offered no explanation, no timeline, no criteria for discharge.

Petros joined in 2012, during what he calls "the quiet years." There was no active war. The border with Ethiopia was frozen, militarised, silent. His unit was deployed to the southern highlands. They dug trenches. They guarded roads. They built a school in one village, a clinic in another. For the first two years, he believed the work mattered. Then the work became maintenance. Then it became waiting.

"You stop thinking about discharge," he said in an interview at Mai Aini camp, conducted through a translator. "You stop asking. The ones who ask too much disappear for a while. They come back different. So you stop asking."

The System's Logic

Eritrea's government has never officially acknowledged that national service is indefinite. When pressed by international bodies, officials cite security threats, the unresolved border dispute with Ethiopia, and the need for vigilance. In a 2023 statement to the African Union, Eritrea's foreign minister described national service as "a civic duty adapted to regional realities." He did not mention duration.

But researchers and defectors describe a different logic. The conscription system functions as a mechanism of population control. It removes young people from the economy, preventing independent livelihoods and political organisation. It provides the government with a captive labour force for infrastructure projects, mining operations, and agricultural schemes — work performed at subsistence wages. It enforces compliance through collective punishment: families of deserters are fined, imprisoned, or stripped of access to government services.

The economic structure is deliberate. Conscripts earn between 500 and 2,000 nakfa per month, depending on rank and assignment. Even at the higher end, this is approximately forty-four U.S. dollars. They cannot refuse assignments. They cannot quit. In practice, they are mobilised labour, deployed wherever the state requires: construction sites, state-owned farms, government offices. Some are assigned to private companies owned by military officers or ruling-party officials, effectively working as unpaid labour for profit-making enterprises.

The Border as Pressure Valve

Petros knew seven men in his unit who had tried to escape. Two were caught within hours and returned to the barracks. They were beaten in front of the others, then taken to a military prison. One came back after four months. The other did not. Three others made it across the border into Sudan. Petros heard later that they had reached Khartoum, then Libya, then drowned in the Mediterranean attempting the crossing to Italy.

The last two made it to Ethiopia. That is the route Petros eventually chose. The Ethiopia-Eritrea border is long, mountainous, and heavily militarised, but it is also the most direct path to asylum. Despite the risks — landmines, shoot-on-sight orders, patrol units — an estimated 3,000 to 5,000 Eritreans cross into Ethiopia every month. Most head for refugee camps in Tigray and Afar regions. From there, some attempt onward migration to Sudan, Kenya, or across the Sahara toward Europe.

The exodus has become structural. UNHCR data shows that Eritrea has produced more refugees per capita than any country in Africa not currently at war. In 2025 alone, 46,000 Eritreans claimed asylum in Europe, the Middle East, and East Africa. The primary reason cited in asylum applications is indefinite military service. Secondary reasons include political persecution, arbitrary detention, and lack of legal recourse.

Regional Destabilisation

The conscription system does not only affect Eritrea. It has become a driver of regional instability, straining host countries, fuelling smuggling networks, and creating a permanent humanitarian crisis that no neighbouring state can resolve. Ethiopia, already managing internal displacement from the Tigray conflict and other regional crises, now hosts five major Eritrean refugee camps. These camps require ongoing international funding, security, and infrastructure. When funding declines — as it did in 2024, when UNHCR faced a 23% budget shortfall — rations are cut, services collapse, and refugees attempt onward migration.

Sudan, embroiled in its own civil conflict, has seen Eritrean refugee camps drawn into militia recruitment. Djibouti, strategically vital and geographically tiny, has closed its borders to new arrivals multiple times. Kenya, farther south, now hosts growing numbers of Eritreans who have made the overland journey through Ethiopia and Somalia. Each transit country faces the same problem: refugees are arriving faster than they can be resettled or repatriated, and Eritrea shows no sign of reforming the system that produces them.

The smuggling economy has grown in parallel. Routes from Eritrea through Sudan to Libya are controlled by networks that charge between $3,000 and $7,000 per person. Many refugees are kidnapped en route and held for ransom in Sudan's desert or Libya's detention centres. Families in Eritrea, already impoverished, pay ransoms that can reach $15,000. Those who cannot pay often disappear.

What the Peace Did Not Change

In 2018, Eritrea and Ethiopia signed a peace agreement, ending two decades of frozen conflict. The agreement was celebrated internationally. Ethiopian Prime Minister Abiy Ahmed received the Nobel Peace Prize the following year. Observers predicted that Eritrea would demobilise its military and begin normalisation. The African Union called it a "new chapter" for the Horn of Africa.

That did not happen. Eritrea did not demobilise. It did not release conscripts. It did not reform national service. Instead, the government maintained the same mobilisation structure, redeployed units to internal security and infrastructure projects, and continued to conscript new cohorts. When Tigray went to war in 2020, Eritrean forces intervened on the Ethiopian federal side — a deployment made possible by the very military structure that conscription sustains.

By 2025, seven years after the peace agreement, refugee outflows from Eritrea were higher than they had been in 2018. The camps in Ethiopia had not emptied. They had grown. Petros was part of that growth. He arrived in Mai Aini camp on March 21, 2024. He was processed, assigned a tent, and given a ration card. He was interviewed by UNHCR officials, who recorded his testimony. He was interviewed again by researchers from the Danish Refugee Council and the Institute for Security Studies in Nairobi. His story was nearly identical to the stories of the 846 others who arrived that month.

Still Waiting

Petros is still in Mai Aini camp. As of March 2026, he has been there for one year. He shares a tent with five other men. He receives a monthly food ration and access to a clinic. He has applied for resettlement to Canada. The average processing time for Eritrean refugees applying to Canada is forty-two months. He does not expect to leave before 2028.

He has not spoken to his family since he crossed the border. He does not know if they were fined or detained. He does not know if his younger brother, now nineteen, has been conscripted. "I think about him all the time," he said. "I think: maybe he will leave too. Maybe he won't. Maybe he'll serve twelve years. Maybe twenty. Maybe his whole life."

The Eritrean government has made no public statement about reforming national service. It has not responded to repeated calls from the African Union, the United Nations, and regional governments to establish clear discharge criteria or timelines. It has not acknowledged the existence of indefinite conscription. In March 2026, a new cohort of 18-year-olds reported to Sawa training camp. They were told, as every cohort before them, that they would serve for eighteen months.

Petros does not believe that. Neither do the young men who cross the border after him, an average of 140 per day, carrying nothing, walking at night, heading toward camps that are already full. The system that holds them is visible, documented, and understood. What remains unknown is how long it will continue, and whether any external pressure can end it. Until it does, the border will remain what it has been for two decades: not a line between countries, but a line between captivity and exile.