Patterns in the Sky

Satellite trains in the sky: what Starlink did to UFO reports

5 min read

Between May 24 and 26, 2019, the National UFO Reporting Center received at least 79 reports of lights moving in a line. In the first three weeks of May, only three such reports had come in, and the first 60 Starlink satellites launched on May 23. To find out how much and for how long the satellite network changed UFO reporting, we went through 67,000 reports from 2014 to 2026.

What we examined

The National UFO Reporting Center (NUFORC) runs the largest public UFO database in the United States. We processed 67,470 reports of sightings that occurred between January 2014 and October 2026, broken down by shape category.

We measured two things each month. One was the number of reports whose summary describes lights moving in a line, a chain or a "train." The other was the number of reports where NUFORC named Starlink as a possible cause in an editor's note or explanation.

The baseline: 14 a month

Between January 2014 and April 2019, 2.9 percent of reports described lights moving in a line. That averaged 14 reports a month, with a high of 38 in December 2017.

Three days, 79 reports

The first 60 Starlink satellites launched on May 23, 2019. The next evening, Dutch satellite tracker Marco Langbroek was already photographing them as they moved across the sky in a line.

Among reports in the light, formation, other, unknown, circle and changing shape categories, 79 described lights in a line between May 24 and 26. In the same categories, only three such reports came from the first 23 days of May. Across the whole database, May 2019 saw 101 such reports, compared with three in April.

The peak: spring 2020

From November 2019 to May 2020, SpaceX launched batch after batch of 60 satellites each. Over those seven months, 5,255 reports came in, and NUFORC named Starlink as a possible cause in 1,246 of them, or 23.7 percent.

The peak came in April 2020. Of the 1,070 sightings that occurred that month, 453, or 42.3 percent, carried a Starlink note. The same month, 230 reports described lights in a line, more than sixteen times the baseline.

Back to normal

On April 28, 2020, SpaceX announced that it would fit its satellites with a sunshade and turn them edge-on to the Sun while raising their orbits, so they would reflect less light. In May 2020, the number of reports describing lights in a line fell to 35.

Between January 2022 and October 2026, 3.8 percent of reports described lights in a line, barely more than before Starlink. NUFORC's Starlink notes fell back to 0.7–3 percent of reports a year.

Two spikes don't fit the picture. In May 2021, 98 reports described lights in a line, but NUFORC flagged Starlink in only 18. December 2021 saw 71 Starlink notes. Our data don't show the cause of either.

Others reached the same conclusion

In 2023, Chase Cockrell (University of Vermont) and Mark Rodeghier and Linda Murphy (Center for UFO Studies) examined NUFORC and MUFON data from 2018 to 2020. They wanted to know whether the pandemic drove the rise in reports in 2020. They found that the increase followed Starlink launches. After removing Starlink-related reports, NUFORC's 2020 total fell below pre-pandemic years.

Data used was provided by the National UFO Reporting Center (nuforc.org) and is used with NUFORC's special permission.

What could the explanation be?

What is certain

  • In the three days after the first Starlink launch, reports of lights in a line jumped in the NUFORC database: 79 in six shape categories, compared with three in the previous 23 days of the month.
  • Between November 2019 and May 2020, NUFORC named Starlink as a possible cause in 23.7 percent of reports, and in 42.3 percent in April 2020.
  • Since 2022, the share of reports describing lights in a line (3.8 percent) has been close to the pre-Starlink level (2.9 percent).

What is not established

  • That every line of lights was Starlink. Such descriptions turned up every month before Starlink, too.
  • That SpaceX's dimming measures caused the decline after 2020. The timing matches, but people getting used to the sight and changes in willingness to report may also have played a part.

Strengths and weaknesses of the method

  • Strength: the spike follows the first launch to the day, and NUFORC's own notes show the same pattern as the keyword filter. A 2023 study of NUFORC and MUFON data reached a similar conclusion.
  • Weakness: the keyword filter is crude. The Starlink notes are NUFORC editors' opinions, and most reporters are in the United States, so the pattern may not hold elsewhere.

Open question

Thousands of Starlink satellites are in orbit today, yet there are fewer reports of lights in a line than in 2020. Whether dimming or people learning to recognize a satellite train matters more could only be settled by a survey or a closer analysis of the report texts.

How we calculated

  • Source: the public database of the National UFO Reporting Center (NUFORC), via its shape-category indexes (nuforc.org/ndx/?id=shape). Downloaded on November 8, 2026.
  • Period: sightings that occurred between January 1, 2014, and October 31, 2026, counted by the month of the sighting.
  • Number of reports: of the database's 162,714 reports, 67,470 fall in this period and belong to a shape category.
  • Lights in a line: reports whose English summary contains the word line, row, string, train, chain or single file. That is 3,085 reports.
  • Starlink notes: reports from 2019 onward where a NUFORC editor's note or explanation mentions Starlink. That is 1,875 reports.
  • Baseline: the monthly average from January 2014 to April 2019 (14 reports, 2.9 percent of all reports).

Data used was provided by the National UFO Reporting Center (nuforc.org) and is used with NUFORC's special permission.

What the data can't tell us

  • Cause and effect. The timing doesn't prove that any single report was Starlink.
  • Exact counts. The keyword filter misses cases described in other words and picks up other things moving in a line, such as lanterns or aircraft. Reports without a shape (6,246) were left out.
  • Whether NUFORC got it right. The Starlink notes are editors' judgments, not measurements, and we don't know whether they applied the same criteria every year.
  • Other countries. Most reporters are in the United States, so the results may not hold for Europe or Hungary.

What was that?

Sources

  1. NUFORC Reports by Shape
  2. NUFORC Reports by Month
  3. NUFORC Reports for Shape Formation
  4. NUFORC Reports for Shape Light
  5. Here’s What SpaceX’s Starlink Satellites Look Like in the Night Sky (Live Science)
  6. SpaceX launches 60 Starlink satellites, aces rocket landing in milestone flight (Space.com)
  7. SpaceX to debut satellite-dimming sunshade on Starlink launch next month (Spaceflight Now)
  8. New study investigates the impact of the COVID-19 pandemic on UFO reporting (EurekAlert!)
  9. Elon Musk’s satellites are muddling UFO sighting statistics, researchers say (KESQ/CTV News)