What Official UFO Investigations Found

Unidentified is a procedural data status, not a conclusion about non-human origin.

The short answer
  • 90% to 97% of reports resolve to known causes. Official datasets (Blue Book, AARO, GEIPAN, Hendry) prove astronomical bodies, aircraft, balloons, drones, and satellites dominate sightings.
  • Venus and bright stars are the #1 trigger. Venus alone caused 35% of Blue Book alerts in Spring 1953; astronomical bodies account for 29% of all misidentifications.
  • Unidentified means data deficit, not alien physics. GEIPAN separates Category C (30% lacking data) from Category D (3% unexplained despite investigation).
  • Trained observers are human. Pilots suffer a 75% error rate on unfamiliar sky lights because featureless skies remove spatial distance references.

For eight decades, government programs, air forces, and scientific institutions have investigated public and military UFO and UAP reports. Across thousands of investigated cases, the base rates are consistent: 90% to 97% of reports resolve to ordinary astronomical, aeronautical, or optical causes.

Before evaluating an unexplained light, check your observation against our master sky field guide. To understand the historical statistics, pilot error rates, and official government findings, use this breakdown of US military investigations, international space agency databases, and Bayesian probability models.

US Official Programs: Project Sign to Pentagon AARO (1947–2026)

Official US government investigation began after WWII and has evolved across several military and intelligence eras:

  • Project Sign (1947–1949, 243 cases) & Project Grudge (1949, 244 cases): Early USAF investigations concluded ~80% of cases were caused by conventional objects, aircraft, or mass hysteria.
  • Project Blue Book (1952–1969, 12,618 cases): The USAF’s longest running program resolved 11,917 cases, leaving 701 (5.5%) unidentified. The 1953 CIA Robertson Panel and 1968 Condon Committee (1,485 pages) concluded UFOs posed no direct national security threat and recommended public education to reduce intelligence clutter.
  • Modern Intelligence Programs (AATIP, UAPTF, AARO):The Pentagon’s All-domain Anomaly Resolution Office (AARO) instituted standardized scientific review:
    • FY2022: 510 total holdings (195 resolved to balloons, drones, or clutter).
    • FY2024: 757 cases analyzed (118 resolved + 174 queued for closure).
    • FY2025: 319 new cases received and 370 total cases closed. Every closed case received a conventional explanation (satellites, balloons, birds, aircraft, drones, or launches). AARO confirmed zero evidence of extraterrestrial origin or breakthrough foreign technology.
  • NASA Independent Study Team (2023): A 16-member scientific panel found no evidence linking UAPs to extraterrestrial sources and recommended multispectral sensor calibration and AI-driven data standards.

International Datasets & Taxonomies: GEIPAN, CRIDOVNI, and MoD

International space agencies and defense ministries maintain transparent, public UAP databases:

Nation / AgencyProgram & DatasetTaxonomic Outcome Breakdown
France (CNES)GEIPAN (3,257 archived cases)Cat A (28% identified), Cat B (39% probably identified), Cat C (30% unidentified due to lack of data), Cat D (3.1% unexplained despite investigation, ~99 cases)
Uruguay (Air Force)CRIDOVNI (1,500+ cases)3% to 4% residual unexplained cases; 96%+ resolved to conventional aircraft or atmospheric events
United Kingdom (MoD)Project Condign (50,000+ pages)Shape taxonomy: Ball (29%), Triangular (8%), ST (7.5%), Oval (3%). Attributed residual to buoyant atmospheric plasmas and meteors
Canada (OCSA)Sky Canada Project (2022)7-point origin matrix: Airborne Clutter, Atmospheric, Human, Industry, Gov Tech, Private Tech, Other

The comparison reveals a universal international baseline: across France, Uruguay, the UK, and the US, genuine high-consistency, high-strangeness unexplained cases (GEIPAN Category D) consistently represent only 2% to 4% of total reports.

Granular Causative Factors: Allan Hendry’s 1,307 Case Benchmark

In 1979, astronomer Allan Hendry published a landmark epidemiological investigation of 1,307 UFO reports for the Center for UFO Studies (CUFOS). Hendry found that 88.6% (1,158 cases) were clearly identified prosaic objects (IFOs), 8.6% (113 cases) were unknowns, and 2.8% were unusable.

Specific Prosaic CauseCUFOS Benchmark (Hendry 1,307 Cases)Modern Pentagon AARO / GEIPAN Dominant Causes
Stars & Planets29% (Venus leading all individual causes)Venus, Sirius, Jupiter (tracked in Cat A/B)
Conventional Aircraft~20% (including night advertising planes)Commercial jets, radial approach landing lights
Meteors & Space Debris9%Bolides, orbital decay re-entries
The Moon1.7% (22 cases)Horizon moon illumination and halo optics
Prank & Weather Balloons1.9% (23 cases)Massive modern surge; leading cause in AARO/UAPTF cases
Drones & Satellites1.8% (24 satellite cases; pre-drone era)Dominant modern cause: Starlink flares, consumer quadcopters
Sensor & Optical Artifacts~1% (Window reflections, mirages)Leading modern military cause: FLIR glare, parallax, rolling shutter
Deliberate Hoaxes<3% (37 cases)CGI, synthetic media, composite photo edits

Temporal Wave Structure & Twilight Physics

Report volumes do not occur uniformly. They spike in “flaps” driven by media contagion, film releases, launch tests, and twilight optical physics:

Sociological Contagion

  • Media coverage triggers “expectant attention”
  • Public scans sky, reporting standard stars/planes
  • Blue Book jumped from 169 (1951) to 1,501 reports in 1952
  • Internet portals increase raw low-quality report submissions

Twilight Geometry Physics

  • Sightings cluster heavily 30–60 min around sunset/sunrise
  • Ground is dark, but high-altitude objects remain sunlit (>50 km)
  • Venus reaches peak brilliancy (−4.9) during twilight
  • Creates high-contrast glowing displays over dark observers

Eyewitness Reliability: Sherif Autokinetic & Aviation Illusions

A common misconception is that trained observers (pilots or police) do not make misidentifications. Allan Hendry’s data showed pilots suffered a 75% error rate when reporting UAPs, while police officers suffered a 94% error rate.

This is a mathematical base-rate effect: pilots spend thousands of hours staring at featureless open skies. Without a background reference frame (trees or buildings), human vision suffers from size-distance invariance failure—it cannot judge the true altitude, physical size, or speed of an isolated point light.

  • Sherif Autokinetic Experiment (1935): Fixed point of light in a dark room. 19 solo subjects perceived a stationary light to drift 20 to 80 cm; groups of 40 formed fabricated consensus.
  • Distance Estimation Failure: In 49 cases where the stimulus was proven to be Venus or a star, witnesses estimated distance at 200 feet to 125 miles—off by hundreds of thousands of miles.
  • Pentagon Video Sensor Resolution:
    • Navy “Go Fast” Video: Proven by trigonometric telemetry analysis (Mick West, NASA 2023 panel, AARO) to be a parallax illusion. A jet at 25,000 ft filmed a slow weather balloon at 13,000 ft drifting at 40 mph over ocean background.
    • Navy “Gimbal” Video: Proven to be a mechanical IR glare rotation. The FLIR camera de-rotation mechanism rotated internal optics, making a static jet exhaust glare appear to physically rotate.

Bayesian Prior Probabilities for an Object-Identification System

To adapt this epidemiological dataset into an automated triage system, base rates must initialize Bayesian prior probabilities:

  • Prosaic Object Prior: Initialized at >0.95 (reflecting GEIPAN, AARO, and CUFOS base rates).
  • Non-Human Craft Prior: Initialized at ~0.00 (reflecting zero verified empirical cases).

Shannon Entropy Reduction Checklist

  1. Fixed Star Alignment: Does the light maintain fixed coordinates relative to background stars? (→ Astronomical body: Venus, Jupiter, or Sirius).
  2. Rhythmic Strobes: Does it flash in a repeating 40–100 FPM sequence? (→ Aircraft or drone navigation lights).
  3. Observer Motion Parallax: Does its movement mirror camera or vehicle motion? (→ Lens flare, window reflection, or parallax).
  4. Twilight Expansion: Is a translucent plume expanding within 60 minutes of sunset/sunrise? (→ Rocket launch exhaust or upper-stage fuel dump).

If a report passes all prosaic filters and enters the 2%–3% residual space, the system must never output “Alien”. Instead, it must trigger a “Category D Anomaly: Human-in-the-Loop Review” flag, escalating the file to human analysts for multispectral sensor cross-referencing.

To test a night-sky photograph for lens flares, rolling shutter, or hot pixels, see our camera photo forensics guide. To check fixed or moving point lights, consult our star-versus-planet guide and moving lights guide.

For the light above you right now, open Astro and point your phone at the sky.

Open the live sky

Frequently Asked Questions

What percentage of UFO reports turn out to be explained?

Across official programs spanning 80 years, 90% to 97% of raw reports resolve to prosaic causes. Project Blue Book (12,618 cases) resolved 94.5%; Allan Hendry's study (1,307 cases) resolved 88.6%; France's GEIPAN (3,257 cases) resolves 97% (leaving ~3% Category D); and the US Pentagon's AARO resolved 100% of its analyzed closed cases in FY2025 to conventional causes.

Has any government investigation admitted UFOs are alien spacecraft?

No. Neither the US Department of Defense (AARO), NASA's 2023 independent study team, France's CNES/GEIPAN, nor the UK Ministry of Defence have found any empirical evidence linking UAP/UFO sightings to non-human intelligence or extraterrestrial technology.

Why do some cases remain listed as 'unidentified'?

In official methodologies, 'unidentified' predominantly represents an information deficit (procedural lack of data), not extraordinary physics. For instance, GEIPAN explicitly separates Category C (30.1% unidentified due to missing sensor/time/optical data) from Category D (3.1% unexplained after exhaustive investigation).

What are the most common things people mistake for UFOs?

Astronomical bodies lead all historical categories (29% in Hendry's benchmark), led by Venus, bright stars (Sirius), and meteors (9%). Conventional aircraft account for ~20%, with weather balloons, commercial drones, satellite flaring (Starlink), and sensor artifacts (infrared glare, parallax) dominating modern reports.

Why do pilots and trained observers also misidentify sky lights?

Trained observers suffer a 75% misidentification error rate (Hendry 1979) because the human visual apparatus lacks background depth cues to judge speed and altitude over open ocean or sky. Pilots are trained to identify military airframes, not atmospheric optics, the autokinetic effect, or satellite glints.

Primary research & datasets

Reference data and official sources cited across this guide:

  1. US DoD AARO — Consolidated FY2024 & FY2025 Annual UAP Reportsaaro.mil
  2. CNES GEIPAN — Official Investigation Database & Statistics (3,257 cases)cnes-geipan.fr
  3. US Air Force — Project Blue Book Historical Case Archives (12,618 cases)af.mil
  4. NASA — Independent Study Team Report on Unidentified Anomalous Phenomena (2023)nasa.gov
  5. Allan Hendry / CUFOS — The UFO Handbook: Epidemiological Study of 1,307 Casescufos.org
  6. UK Ministry of Defence — Project Condign UAP Study Archivesnationalarchives.gov.uk