MU5735 records show fuel switches moved to cutoff; no final report
NTSB records show fuel-switch and flight-control changes on China Eastern flight MU5735, and document recovered cockpit audio. The files do not determine the crash’s cause. SharpPost found no public final report in its review through October 5, 2026.
Updated and corrected October 5, 2026. This revision checks the investigation’s public status and corrects recording duration, data completeness and disclosure rules. The original publication date is unchanged.
Recorder files released in May 2026 under the US Freedom of Information Act show both engine fuel switches on China Eastern flight MU5735 changing from RUN to CUTOFF, followed by changes in autopilot and flight-control parameters. The records were produced during US assistance to China’s investigation. They are not a determination of the crash’s cause. SharpPost’s review of publicly available investigation material through October 5 found no published final accident report.
The release includes a Combined Download Report dated July 1, 2022, flight data and correspondence. Flightradar24’s account of the disclosure says the NTSB received the request in January 2026 and released data from the accident flight and the preceding flight. Approximately 23 seconds elapsed between the fuel-switch changes and the end of the recording. The released files cover a much longer period.
Fuel switches and flight controls
Pages 25–27 of the report plot basic aircraft parameters, control positions and control forces. The aircraft was still near cruising altitude when the fuel-switch states changed. Autopilot-related states then changed, alongside substantial pitch, roll and control-force variations. The final altitude readings are around 26,000 feet. The plots are marked “PRELIMINARY” and use a recording-time axis, rather than directly displaying local clock time.
| Parameter | Recorded observation | Not established by this trace alone |
|---|---|---|
| Fuel switches | Both change from RUN to CUTOFF at closely spaced times | Who operated them or why |
| Autopilot | CMD A changes from selected to not selected; warning states change | The full explanation for disengagement or events inside the cockpit |
| Flight controls | Column position, pitch forces and roll parameters change | The identity or intent of an individual, or whether a struggle occurred |
| Recording endpoint | Pressure altitude around 26,000 feet; recording ends before impact | The continuous flight history during the missing interval |
The traces help reconstruct the sequence of switch changes and aircraft responses. They do not identify an operator. In May 2022, The Wall Street Journal cited people familiar with a preliminary US assessment pointing to deliberate control inputs. CNN’s account of that report also carried China Eastern’s response that no evidence had yet established whether there were problems with the aircraft. Neither the reported assessment nor the airline’s response was a final investigation finding.
Page 23 explains why the FDR stopped before impact: after the fuel switches moved to cutoff, engine speeds fell below the level needed to keep the generators online. The FDR has no battery backup and lost power. The CVR does have backup power, which the report says can support at least 10 minutes of recording after generator power is lost. That equipment capability does not establish how long any post-power-loss audio lasted in this accident.
What was recovered from the two recorders
The report lists four NTSB specialists and six CAAC participants. NTSB personnel performed the recovery work, overseen and recorded by the Chinese delegation. Recorder specialist Charles Cates authored the report, which describes damaged circuitry, connector repairs, chip readouts and timing corrections. It does not assign responsibility for the crash.
Five of the flight data recorder’s six memory chips could be read; chip U2 could not be recovered. Because the data were written across all six chips, the reconstructed stream contains about 6.5 seconds of valid data followed by a 1.3-second gap. Five readable chips does not mean that every parameter was recovered with 83% completeness. Engineers also corrected erroneous timing gaps. The final 12 minutes of the accident flight’s recorded data were time-corrected, and approximately 150 parameters were validated.
| Item | Flight data recorder — FDR | Cockpit voice recorder — CVR |
|---|---|---|
| Model and serial | Honeywell HFR5-D; FDR-02952 | Honeywell HFR5-V; CVR-04014 |
| Recovery | Five of six chips readable; periodic data gaps | All four channels rated Excellent; brief losses in crew channels |
| Duration | Final 12 recorded minutes time-corrected; report plots show the final 90 seconds | Three audio-panel channels about 120 minutes each; area microphone about 180 minutes |
| Content and units | 512 twelve-bit data words per second—not 512 distinct parameters | Captain, first officer and observer audio panels; cockpit area microphone |
| Copies and transfer | Files given to CAAC; NTSB retained FDR copies | Audio and files capable of generating it given to CAAC; NTSB retained none |
| Public material | Recovery report, parameter plots and FDR data | Recovery methods, duration and quality—not the actual cockpit conversations |
Early CVR readouts contained stuttering, echoes and digital noise. After connector repairs, all four channels were rated “Excellent”. The report also identifies occasional short data losses in the crew channels and timing drift between channels. It says the accident was captured; it does not publish what the crew said or describe the content of their conversations or cockpit sounds.
Page 14 states that the NTSB retained no CVR audio or raw or intermediate download files that could generate audio. It kept photographs, scans and microscopic images. FDR copies were retained so the NTSB could continue assisting CAAC with the flight history and sequence of events.
The investigation’s public record
MU5735 crashed near Wuzhou in southern China on March 21, 2022 while flying from Kunming to Guangzhou. CAAC confirmed that all 123 passengers and nine crew members died. Its April 20, 2022 bulletin on the preliminary report described the aircraft, crew, weather and damaged recorders. The FOIA release was therefore not the first official information published in four years.
| Date | Event or document | Scope |
|---|---|---|
| March 21, 2022 | MU5735 crash | 132 deaths; China leads the investigation |
| April 20, 2022 | CAAC preliminary-report bulletin | Initial information; no determination of cause |
| July 1, 2022 | NTSB Combined Download Report completed | Recorder recovery and processing; not a final report |
| March 20, 2024 | CAAC progress bulletin | Wreckage, operations, airworthiness and support checks; analysis continuing |
| January–May 2026 | FOIA request and release | FDR data, recovery report and some correspondence made public |
| October 5, 2026 | Cutoff for this revision | No public final report or new determination of cause found |
CAAC’s March 20, 2024 bulletin reported no abnormalities in specified checks covering qualifications, aircraft documents, loading and weather, while laboratory work and analysis continued. Those individual findings do not explain why the aircraft descended abruptly or exclude every possible technical cause.
Flightradar24 reported in its June 2025 update that Chinese authorities withheld annual investigation-progress material, citing risks to national security and social stability. A progress report is not the same document as a final accident report. That response does not establish that a completed final report exists but has been withheld.
| Question | Rule | Application here |
|---|---|---|
| Final-report timing | ICAO Annex 13 calls for release as soon as possible, if possible within 12 months; annual interim statements if unfinished | The earlier article’s blanket three-year final-report deadline was wrong |
| US-held foreign-investigation records | 49 USC §1114(f): release when the investigating country issues its final report or two years after the accident, whichever is earlier | Disclosure does not transfer leadership of the investigation to the NTSB |
| Cockpit recordings | §1114(c) separately restricts public release of CVR recordings and sets conditions for relevant transcripts | General disclosure of foreign-investigation records does not require release of the audio |
| Custody of this CVR | NTSB says it gave the audio and recoverable source files to CAAC and kept no copies | Possession of files and legal authority to disclose them are separate issues |
China conducts the investigation as the state where the accident occurred. The United States participates as the aircraft’s state of design and manufacture, with the NTSB providing technical assistance. The FOIA records cannot replace the investigating authority’s assessment of wreckage, operating records, flight data and cockpit audio.
The public files establish that recorder data were recovered and that fuel-switch and control parameters changed. They do not explain who carried out the relevant actions or why. A formal account of the full sequence, causes, contributing factors and safety recommendations remains absent from the public material reviewed for this update.
Corrections: The earlier article misstated the scope of the “23 seconds”, treated Excellent audio quality as complete recovery without gaps, and misstated ICAO reporting deadlines. Its claim of a May NTSB hearing on MU5735 has also been removed: the verified May hearing concerned UPS flight 2976. The technical report is available as an archived NTSB document; the official disclosure portal is the NTSB FOIA Reading Room.
You read this far. You're not here for noise.
SharpPost delivers one weekly deep dive on geopolitics, finance, and tech — decoded for readers who want signal. Carefully selected, never noisy.