Lost Lessons Due to Air Raid Alerts: How Many — and What Can Be Done?

On September 1, 2024, Ukraine’s Ministry of Education rolled out a new policy called “Offline School.” Its goal is to bring students back into classrooms after years of forced remote learning. The policy responds to serious declines in the quality of education and student socialization caused by the shift to online learning in 2020. These setbacks have been confirmed by PISA results and other studies showing the negative impact of remote education on children’s behavior and well-being.

The comprehensive “Offline School” policy is aimed at returning children to offline schooling. Under the Ministry of Education’s order, schools that can provide students with access to shelters—on-site or within 500 meters—must organize instruction in either offline or blended format. The order does not apply to regions facing more severe security risks. 

Since its rollout, the policy has shown promising results: in 2023, 18% of all students were studying remotely; by 2024, that number reduced to 11.7%. This means that nearly 200,000 children have at least partially returned to offline schooling.

Since schools are responsible for student safety regardless of the study format, lessons must be paused during air raid alerts. In the case of offline schooling, students must go to a shelter where, according to the Ministry of Education letter sent to local educational departments, classes may continue if the shelter is properly equipped. This provides offline schooling a key advantage over remote formats, which are simply interrupted. 

However, most schools cannot continue classes in shelters because shelters are usually small basements with limited space. To accommodate all students, some schools have had to introduce second shifts. As a result, shelters continue to serve primarily as safety zones.

For the vast majority of schools, this means that air raid alerts stop the lessons until the all-clear is given. Frequent alerts lead to significant learning losses that must be properly assessed. In this article, we present a methodology for calculating potentially lost lessons, examine region-level data, and offer recommendations for reducing foregone time in class. 

Methodology 

According to data obtained from the Institute of Educational Analytics, as of September 1, 2024, 267,000 of Ukraine’s nearly 3.7 million students attend school in the second shift (this number may be imprecise, as some schools operate in two or even three shifts due to limited shelter space, and this is not always reflected in official reports). This means that about 90% of Ukrainian students attend the first shift, with classes typically starting at 8:30 a.m. and the eighth lesson ending at 4:00 p.m. As a result, when calculating lost classroom time, the first shift should be used as the standard reference.

This means that even in upper secondary school, students typically have up to seven lessons a day, with eight as the exception. Air raid alerts that begin before the start of the school day can lead to cancellations, postponements, or shortened lessons due to transportation difficulties and uncertainty about the alert’s duration. As a result, the highest risk of lesson disruptions or cancellations comes from alerts that occur between 8:00 a.m. and 3:00 p.m.

Data on the location, timing, and duration of air raid alerts comes from a dataset published on GitHub. The data is sourced and updated from official channels via eAlert. 

Although air raid alerts are generally shorter during school hours, calculating learning losses requires factoring in not just the duration of the alert itself but also the time needed to restart the educational process—at minimum, the time it takes for students to return from the shelter to the classroom. In many cases, a paused lesson cannot simply be resumed. Teachers may try to continue during the break period, but if the alert is lengthy, that time is often insufficient. To standardize the approach, we assume that any alert lasting between 5 and 60 minutes and occurring between 8:00 a.m. and 3:00 p.m. results in the loss of one full lesson.

This means that if an alert lasts less than five minutes during the designated time window, it is not counted. If it lasts between five and 59 minutes, for example, 20 minutes, it cancels [20/60+1] = 1 lesson. Alerts that last more than 60 minutes result in the cancellation of two or more classes.

Students in the 1-4th grades have less schooling time than those in middle and upper grades. This applies not only to the number of lessons but also to their duration: first graders have 35-minute lessons, while students in grades 2–4 have 40-minute lessons, while the standard time is 45 minutes. Their school day ends by 1:00 p.m. As a result, the time window in which air raid alerts can disrupt classes is shorter for primary school students. 

Since the analysis is conducted month by month, it is necessary to estimate the typical number of lessons per month. For primary school students (grades 1–4), that range is approximately 80–100 lessons. For middle school (grades 5–9) and upper school students (grades 10–12), it is 120–140 lessons per month. Air raid alerts that occur on weekends or during school holidays are excluded from the calculations. In January 2025, for example, there were only 15 school days (January 13th to 31st), so the number of missed lessons was lower too (February 2025 had 20 school days). Given the variation in teaching time, alert frequency, and vacation schedules in different months, the average from September 2024 to February 2025 offers the best estimate.

Our analysis focuses primarily on oblast-wide alerts rather than rayon-level ones, as rayon-level alerts are not used everywhere and do not affect all the students. In some oblasts, there are communities classified as facing an “unmanageable level of risk,” such as the southern part of the Nikopol district in the Dnipropetrovsk region.

 This analysis only briefly addresses the regions where active hostilities are ongoing—Kherson, Zaporizhzhia, Donetsk, and Kharkiv. According to Ministry of Education data from December 2024, online learning remains the dominant format in these areas. While online lessons are also paused during air raid alerts, students can study independently in the asynchronous learning format that allows students to access educational materials at their convenience.

Findings

Students in the Sumy region are the most affected by air raid alerts in terms of lost classroom time

The median duration of an air raid alert in Ukraine announced on weekdays between 8:00 a.m. and 3:00 p.m. from September 2024 to February 2025 was 33 minutes. This means that in most cases, alerts did not last the full 45 minutes of a typical lesson in grades 5–9. Still, when accounting for the time needed to lead students to shelters and bring them back, an entire lesson could be lost. On days with multiple alerts, monthly learning losses can reach up to 100 lessons (see Figure 1).

The Sumy region significantly differs from the rest: on average, air raid alerts led to the loss of 101 lessons per month there, which is about 72–84% of all scheduled classes. This is explained by the ongoing threat of artillery shelling in the area.

The Poltava and Dnipropetrovsk regions also have visibly higher losses than the majority of other regions, with students missing on average 49 and 44 lessons per month, respectively. In Poltava’s case, this can be explained by its proximity to the Sumy and Kharkiv regions. For the Dnipropetrovsk region, the figure reflects its shared border with all regions currently experiencing active hostilities. At the same time, the number of missed lessons in the Dnipropetrovsk region could have been even higher but some air raid alerts were issued only in the Nikopol district, which we excluded from the calculations.

The Chernihiv region ranks fourth in terms of potential educational losses, with an average of 35 missed lessons per month. It is followed by the Kirovohrad, Cherkasy, and Kyiv regions, each averaging 27 missed lessons monthly. This suggests that the most significant educational disruptions occurred in Ukraine’s northern and central regions.

The southern regions experienced relatively moderate losses: 18 lessons per month in Odesa, 19 in Mykolaiv, and even in Kherson—despite ongoing hostilities—only 20. A similar figure was recorded in the Zhytomyr region, with 15 missed lessons per month. The lowest levels of educational loss were observed in Kyiv city and the Vinnytsia region, with 11 and 10 missed lessons per month, respectively, as well as in western Ukraine, where on average fewer than 10 lessons per month are missed. 

A similar pattern is observed in primary schools, although with some peculiarities. Most notably, the gap between the Sumy region and other conflict-affected regions is smaller in the case of primary grades. In these areas, the share of disrupted lessons can range from 72% to as high as 90% of the total. In the Chernihiv region, the situation is somewhat better— at par with the Kirovohrad and Kyiv regions. In Zhytomyr, it is worse, with an average comparable to that of the Odesa region. Meanwhile, in Kyiv, fewer than 10% of primary school lessons are suspended and/or canceled.

In absolute numbers, the Dnipropetrovsk region experiences the highest number of missed lessons—approximately 61,600 per month. This is twice as many as in the Sumy and Poltava regions

The war has led to significant displacement of children to safer regions, resulting in concentration of student populations in certain areas. By the end of 2024, the largest student populations were in Kyiv (30,100), the Dnipropetrovsk region (28,000), and the Lviv region (27,700), together accounting for 24.9% of all students in Ukraine.

Even now, however, regions with ongoing hostilities—namely Kharkiv (15,800 students), Zaporizhzhia (9,600), Donetsk (5,500), Kherson (3,300), and Luhansk (1,200)—account for 10.3% of all students. The three regions facing the greatest educational disruptions (Dnipropetrovsk, Poltava with 12,600 students, and Sumy with 7,900 students) together are home to 14.1% of Ukraine’s student population. Meanwhile, 60.8% of students live in regions where the average number of missed lessons for grades 5–12 does not exceed 15 per month. The remaining 14.8% reside in areas with an average monthly loss of 15–30 lessons.

Assuming an average class size of about 20 students, the estimated number of missed lessons per month from September 2024 to February 2025 was approximately 61,600 in the Dnipropetrovsk region, 39,900 in the Sumy region, and 30,900 in the Poltava region. In Kyiv, the area with the highest student population, around 16,500 lessons were lost per month, while in the Lviv region, only about 6,900. Although these figures are approximate, they clearly illustrate the scale of the problem.

District-level alerts reduce the number of missed lessons

At the beginning of 2025, considering the Cherkasy region’s large area and geographic position, a pilot project was launched to issue air raid alerts at the district (rayon) level. Issuing alerts across the entire region when only a small part is under threat was deemed inefficient. As shown in the alert map for the Cherkasy region (Figure 3), western districts experienced alerts far less frequently, significantly easing the burden on schools.

In January 2025, only eight lessons were missed due to region-wide alerts, with all other disruptions resulting from district-level alerts. The lowest number was recorded in the Zvenyhorod district (one lesson) and the highest in the Zolotonosha district (11 lessons). Even the combined total of 19 missed lessons in the Zolotonosha district (11 district + 8 region) is still lower than the 23 lessons that would have been lost had all alerts been region-wide. This means district-level alerts helped Cherkasy schools preserve between 4 and 14 lessons in January alone—up to 12% of the monthly total.

Another clear example of the benefits of district-level alerts is the significant contrast in educational losses between the city of Kyiv and the Kyiv oblast. On average, the oblast has 16 more missed lessons per month in the middle and senior school and 13 more in primary school, compared to Kyiv. Although Kyiv is administratively separate from the region, this difference illustrates how districts like Vyshhorod, located north of the capital, could avoid declaring alerts triggered by threats originating in more distant southern districts—such as Fastiv, Obukhiv, or Bila Tserkva. Currently, we see situations where Kyiv, which is closer to the threat, does not declare an alert, while more distant towns in the region do.

In addition to the Cherkasy region, district-level alerts have also been used in the Zhytomyr region. However, during the analysed period, most of these alerts occurred at night or on weekends and therefore had little impact on the educational process.

In some regions near the border or active combat zones, alerts are already issued specifically for high-risk districts—for example, the Nikopol district in the Dnipropetrovsk region and the Pokrovsk district in the Donetsk region. However, most other threats still trigger region-wide alerts. Since February 2025, the Kharkiv region has implemented a separate air raid alert system that distinguishes between the city of Kharkiv and the rest of the region. Discussions are currently underway about expanding this system to include district-level alerts.

Conclusions and recommendations

The case for expanding district-level alerts

The examples of the Cherkasy region, as well as Kyiv city and Kyiv region, demonstrate that district-level alerts can significantly reduce learning losses. This approach is particularly relevant in areas with high rates of lesson suspensions, such as the Sumy, Poltava, Dnipropetrovsk, and Chernihiv regions. In western Ukraine, alerts are more frequently triggered by nationwide threats (e.g. MiG launches), whereas in Poltava, threats may arise from multiple directions, affecting only specific districts.

It is important to remember that the Air Force does not consider a differentiated air raid alert system suitable for threats involving ballistic missiles, cruise missiles, artillery, or multiple launch rocket systems (MLRS). Therefore, the most appropriate application of district-level alerts is in response to UAV (drone) threats. The State Emergency Service, together with representatives of the regional military administration, has referred to the Air Force Command regarding the potential implementation of a differentiated alert system. Its feasibility depends on two key factors: (1) the Air Force’s ability to classify threats, accurately forecast their trajectory in a timely manner, and issue a danger alert for a specific district; and (2) the regional administration’s ability to promptly receive this alert and provide the information to lower administrative units. 

Adapting shelters in regions where air raid alerts more frequently disrupt lessons to allow the continuation of the learning process

However, in regions where alerts more frequently interrupt lessons, transitioning to district-level alerts alone is unlikely to significantly improve the situation. That is why it is crucial to adapt schools to allow the continuation of learning in shelters during air raid alerts. In regions with the highest average number of alerts—Sumy, Dnipropetrovsk, and Poltava— equipping shelters in large cities for educational use should be prioritized, as these areas have the highest concentration of students and, consequently, the greatest learning losses due to alerts.

At the beginning of 2024, the Ministry of Education launched the My Fortress project, which provides standard design and cost estimate templates for underground schools, with the flexibility to adapt them to the specific needs of each institution. Using these templates, school principals can upload their project to the DREAM system and, if they meet certain criteria, such as at least 200 students, confirmed community co-financing of 10% of the total project cost, and completion of construction by the end of 2024, they may apply for the state subsidy.

According to Ivan Fedorov, head of the Zaporizhzhya Regional Military Administration, a standard underground school with an area of 1,300 square meters designed to accommodate approximately 500 students per shift costs around UAH 110-115 million. However, the final cost may vary depending on factors such as soil type and groundwater levels.

As of April 2025, according to the Ministry of Education, construction of 119 underground schools is scheduled for completion by the end of the year in the Kharkiv, Zaporizhzhia, Mykolaiv, Dnipropetrovsk (Kryvyi Rih), Kherson, Chernihiv, and Sumy regions, funded by both state and local budgets. In these regions, excluding schools located in areas deemed to have an unmanageable level of risk, there are 2,912 schools, of which 1,346 have more than 200 students. This means that by September 2025, underground classrooms will be built for fewer than 10% of the larger schools.

These figures do not include the Poltava region, which also ranks among the leaders in the number of air raid alerts. It adds another 173 schools with more than 200 students. As a result, the initiative will cover only a small fraction of educational institutions.

In some regions, educational losses due to air raid alerts may prove greater than those experienced during the pandemic and the shift to remote learning. This poses a serious challenge for Ukraine’s education system, underscoring the importance of acknowledging these losses and taking active steps to minimize and compensate for them.

The authors extend their special thanks to Valentyn Hatsko for his assistance in developing the methodology and initial calculation models, as well as to Yehor Stadnyi for his contributions.