visua.no Ukraine in open data

Ukraine in open data. Visual, documented, auto-updated.

Air war

Russia launched at least 6,649 long-range attack drones in September

Number of long-range attack drones launched by Russia and neutralised by Ukrainian air defence.

Reuse in Datawrapper Download a table Download an image Embed code

What you should know about this chart

  • The chart shows how many long-range one-way attack drones Russia launched at Ukraine and how many of those were neutralised from October 2022 until the last complete month, based on Ukrainian Air Force reports on the official Telegram channel. The width of the red band on the chart shows the number of drones that went through air defence.
  • The series covers Shahed-type long-range attack drones and the cheaper versions and decoys like Gerbera or Parodiya flown alongside them. From 29 September 2026 the series also includes the jet-powered Geran-4 and Geran-5. The Air Force's figures for these may not be complete: in every report during the first month of reporting, the number launched equals the number destroyed, which may mean that only the jets brought down are reported. Reconnaissance drones and short-range loitering munitions – Orlan, ZALA, Supercam, Lancet or Molniya – are excluded.
  • The number of drones is a low-end estimate, as it draws on reports about the major missile and drone attacks. Individual launches outside the big overnight attacks may not be reported at all. Air defence is not the Air Force alone. Units under other commands bring down drones and missiles too, and their tallies are not in these figures.
  • "Neutralised" means the drone did not reach its target – destroyed or knocked off course by electronic warfare. It is deliberately not "shot down": a drone that simply malfunctioned counts here too. Drones still airborne when the report was filed are excluded, because the outcome was unresolved rather than negative.
  • The number of Shahed-type drones used in attacks has steadily risen since 2023. It peaked in May 2026 with 8,207 drones. It fell in June and July, then began rising again in August. In spring 2026, jet drones began appearing in the attacks, and their numbers have been rising. Between October 2022 and September 2026, the average interception (or neutralisation) rate was 87.3%. The rate has stayed largely unchanged despite launch volumes rising roughly thirty-fold.

About the source

Ukrainian Air Force reports, compiled by Petro Ivaniuk

kaggle.com/datasets/piterfm/massive-missile-attacks-on-ukraine

What it is — A record of Russian air attacks on Ukraine: what was launched, of which type, from where, and how much of it did not reach a target. It runs from October 2022 to the present.

Where the numbers come from — Following every major air attack, the Ukrainian Air Force posts a report on its official Telegram-channeldetailing the weapons detected, tracked, intercepted, or knocked off by electronic warfare. Those reports are the original source. Petro Ivaniuk, a Ukrainian data scientist, compiles them into a dataset, which is openly available on Kaggle.

What it counts — It counts missiles and drones that were detected, tracked, and neutralised by Ukrainian Air Force units. The numbers of drones and missiles are a low-end estimate, as it draws on reports about the major missile and drone attacks. Individual launches outside the big overnight attacks may not be reported at all. Air defence is not the Air Force alone. Units under other commands bring down drones and missiles too, and their tallies are not in these figures.

How accurate is it — It is a lower-end estimate. The numbers are definitely imprecise in most cases, but they are useful to see the overall trend. On the one hand, the Ukrainian Air Force has an incentive to inflate the number of intercepted targets. On the other hand, they have an interest in accurate statistics because they rely on weapons supplies from Western partners. The public perception of the number of strikes among Ukrainians also plays a role. A good illustration is that in early August, due to a shortage of interceptor missiles for Patriot systems, the Air Force published a report showing zero interceptions out of 24 launched ballistic missiles. Ultimately, they made the decision to stop publishing statistics regarding ballistic missiles altogether.

Retrieved on
2026-09-30
Coverage
October 2022 — present
Update cadence
Weekly
Citation
Ivaniuk, P. Massive Missile Attacks on Ukraine [dataset]. Kaggle. Compiled from Ukrainian Air Force reports.

How we processed the data

  1. 1 Load the published dataset We download the current dataset and check its columns. If the structure has changed (it did a few times over the years), the processing is stopped.
  2. 2 Group weapons into types Each row names a specific model of a drone or missile. We group those into ballistic, cruise (subsonic), cruise (supersonic), aeroballistic, Shahed-type drones and jet-powered drones. The drone chart counts the last two together. A model we have no rule for stops the load rather than being quietly dropped.
  3. 3 Set aside the tactical fleet Reconnaissance drones, short-range loitering munitions, and glide bombs are excluded. These charts are about long-range strike.
  4. 4 Add up by month Launches are summed per calendar month. The current month is left out until it ends. An attack is counted in the month it began. The weapons used in the attack that started at 18:00 on 31 July 2026 and ended at 09:30 the next day are counted in July statistics.

Changelog

Editorial changes

  • 2026-10-15

    From 29 September 2026 the chart also counts jet-powered drones, the Geran-4 and Geran-5, which the source began listing that week. They are counted together with the Shahed-type drones.

  • 2026-10-15

    For the first two weeks of October, the chart showed 5,318 drones for September 2026, because the last five days of the month had not yet been loaded. The complete figure, including the first jet-powered drones, is 6,649.

Data dictionary

Columns in the downloadable CSV — the exact output of this chart’s processed view. Periods without a usable upstream snapshot are absent rows, never interpolated or carried forward.

ColumnTypeUnit / formatDescription
month text YYYY-MM The completed calendar month. A month appears only once the data covers it to the last day, so the current month, and any month the source has not finished reporting, is excluded.
Launched number drones Long-range attack drones detected and tracked over Ukraine that month: Shahed-type drones and decoys, and from 29 September 2026 the jet-powered Geran-4 and Geran-5.
Neutralised number drones Did not reach the target: destroyed or knocked off course by electronic warfare. Excludes drones still airborne when the report was filed.

The full dataset carries more fields than we use, including launch site, target and per-attack notes. The columns above are the ones these charts are built on.

Reuse this work

You’re free to

  • Republish the chart on your own site, in print, in a paper.
  • Restyle the chart — change colours, fonts, layout.
  • Remix the data with other sources to build something new.
  • Use the chart commercially. <mark>Make sure you check the terms of primary data source — {source} publishes under its own terms.</mark>

The only ask

  • Credit the original source (Ukrainian Air Force) and the processed dataset (this site).
  • Link back to this page when republishing online.
  • Don’t imply Ukrainian Air Force or this site endorses your republication.

All charts and processed datasets on this site are released under Creative Commons Attribution 4.0 (CC-BY 4.0). The underlying source data is licensed by its respective provider — Ukrainian Air Force publishes under its own terms.

Citations

Please cite both this processed dataset and, most importantly, the original Ukrainian Air Force source. The Ukrainian Air Force's work should be recognised. Two-part citations look like this:

In-text

Ukrainian Air Force reports, compiled by Petro Ivaniuk (2026), processed by visua.no

Full citation

Ukrainian Air Force reports, compiled by Petro Ivaniuk (2026). "How many long-range attack drones Russia launches every month" [chart]. Processed by Pavlo Krasnomovets, visua.no. Data as of 30 September 2026. https://visua.no/en/charts/attacks-ua-drones/

BibTeX

@misc{ukrainianairforcereportscompiledbypetroivaniuk2026attacksuadrones,
  author       = {Ukrainian Air Force reports, compiled by Petro Ivaniuk},
  title        = {How many long-range attack drones Russia launches every month},
  year         = {2026},
  note         = {Processed by Pavlo Krasnomovets, visua.no},
  howpublished = {\url{https://visua.no/en/charts/attacks-ua-drones/}},
  urldate      = {2026-09-30}
}

Download

Everything used to produce this chart. The CSV and PNG come straight from Datawrapper; the processed series and source data are linked from the pipeline repository.

Embed code — responsive iframe

Iframe

<iframe title="Russia launched at least 6,649 long-range attack drones in September" aria-label="Line chart" id="datawrapper-chart-UVG2R" src="https://datawrapper.dwcdn.net/UVG2R/" scrolling="no" frameborder="0" loading="lazy" style="width: 0; min-width: 100% !important; border: none;" height="400" data-external="1"></iframe>

Use the unversioned dwcdn.net/UVG2R/ path — the daily refresh always serves the latest version. Full snippet from Datawrapper →

For scripts & agents

Every chart is built to have .csv and .meta.json siblings at a stable URL — no auth, no rate limits for reasonable use. Pages are plain HTML, so the methodology above is readable without JavaScript.

  • Dataset (CSV)

    https://visua.no/data/attacks-ua-drones.csv

    Latest full series. Stable URL — re-fetch daily, content updates in place.

  • Metadata (JSON)

    https://visua.no/data/attacks-ua-drones.meta.json

    Schema, source, processing steps, key points, license. Machine-readable version of this page.

  • Site index for agents

    https://visua.no/llms.txt

    Plain-text index of every chart, dataset URL, and methodology page on the site.

Questions, corrections, source ideas

Using this chart in a story or a paper and something’s unclear? Spotted an error? Know a data source worth tracking? I read (almost) everything — usually same-day.

Email me questions & ideas