NTSB CAROL · Event
Event MIA95IA187
Aircraft involved
Probable cause & findings
CONTAMINATION OF THE WIRING HARNESS AND ELECTRICAL CONNECTORS TO THE E-1-2 AND E-1-3 SHELF DISCONNECT BRACKETS IN THE ELECTRONICS COMPARTMENT, WHICH RESULTED IN AN ELECTRICAL FAULT AND AN UNCOMMANDED ROLL MALFUNCTION OF THE AUTOPILOT. A FACTOR RELATING TO THE INCIDENT WAS: INADEQUATE MAINTENANCE INSPECTION OF THE AIRCRAFT BY COMPANY MAINTENANCE PERSONNEL.
Factual narrative
On July 18, 1995, about 1630 eastern daylight time, a Boeing 737-3B7, N390US, registered to and operated by USAIR, Inc., as flight 855, a 14 CFR Part 121 scheduled domestic passenger service from Pittsburgh, Pennsylvania, to Orlando, Florida, experienced an uncontrolled roll to the left during descent for landing. Visual meteorological conditions prevailed at the time and an instrument flight rules flight plan was filed. The aircraft was not damaged and the airline transport-rated pilot, first officer, 3 flight attendants, and 115 passengers were not injured. The flight originated from Pittsburgh, on July 18, 1995, about 1504. The captain stated that about 65 nautical miles north of Ormond Beach, Florida, the flight was descending through 28,000 feet. The "A" autopilot system was on longitudinal navigation and vertical speed was selected for pitch control. The aircraft suddenly rolled hard to the left to about 25 degrees of bank angle. He overrode the roll with aileron control and returned the aircraft to wings level. The aircraft again rolled to the left and he disengaged the autopilot and yaw dampener. He reengaged the yaw dampener, left the autopilot off, and continued the flight to Orlando, without further incident. Examination of the aircraft showed the connector plugs and wiring harness to the E-1-2 and E-1-3 shelf disconnect brackets in the electronics compartment were contaminated with blue residue and water. The flight control computers and flight management computer mounts on these shelves. Evidence indicated the forward lavatory located above the electronics compartment had leaked at one time but was not currently leaking. Additionally, the main cabin door drain was found to not be properly sealed and water was leaking from the drain and dripping on to the wiring harness and migrating into the connectors. The flight control computer, and roll actuators were removed from the aircraft and tested. No failures which would cause an uncommanded roll were noted. For additional information see attached Boeing Commercial Airplane Company report, the Honeywell test report, and the FAA test report. THE FLIGHT EXPERIENCED AN UNCOMMANDED ROLL TO THE LEFT DURING DESCENT FOR LANDING. THE FLIGHT CREW DISENGAGED THE AUTOPILOT AND YAW DAMPER, BUT REENGAGED THE YAW DAMPER SINCE THERE WAS NO APPARENT YAW PROBLEM. THE DESCENT WAS CONTINUED AND A LANDING WAS MADE WITHOUT FURTHER INCIDENT. EXAMINTATION REVEALED THE CONNECTOR PLUGS AND WIRING HARNESS TO THE E-1-2 AND E-1-3 SHELF DISCONNECT BRACKETS IN THE ELECTRONICS COMPARTMENT WERE CONTAMINATED WITH WATER AND A BLUE RESIDUE. THE SOURCES OF CONTAMINATION WERE WATER FROM A CURRENT MAIN CABIN DOOR LEAK AND BLUE RESIDUE FROM A PREVIOUS LEAK OF THE FORWARD LAVATORY. Source: NTSB Aviation Accident Database (Pre-2008 Archive) Retrieved: 2026-02-12
Verbatim from NTSB's published report. Source file
NTSB_1995_MIA95IA187.txt.
Findings + structured fields enriched from FAA avall.mdb.
Full investigation docket on
data.ntsb.gov ↗.
Beyond the agency record
Search this event elsewhere.
Pre-filled searches into the sources where news + community discussion of aviation events lives. External sources are reported, not agency. Treat them as signal that something happened, not as fact about what happened.
Entity-clustered aviation events in the press — last 24 hr + 30-day archive.
Official agency record + docket.
Investigative docket: factual reports, photos, transcripts.
Long-running aviation incident database (Flight Safety Foundation).
Community NTSB synthesis blog — often has photos and witness reports.
Gold-standard aviation incident blog.
Aviation industry news search.
GA pilot forum — informed but rumor-prone.
GA pilot subreddit search.
Tail-number page — flight history (free tier limited).
AOPA Air Safety Institute search.
Mainstream press coverage. Recent events only.
Privacy-preserving news search.
External links open in a new tab. We don't ingest their content; we deep-link search queries.
Related research
What the literature says.
Academic papers and agency reports matching this event's aircraft type or causal vocabulary (maintenance, autopilot). Sourced from NASA NTRS, NTSB Safety Studies, FAA CAMI, AOPA Air Safety Institute, Embry-Riddle Scholarly Commons, arXiv, and the Semantic Scholar academic graph.
- Embry-Riddle Scholarly Commons 2026 · Journal article (IJAAA)
From Reactive to Predictive: A hybrid Trust-Mediated Adoption Framework for Data-Driven Maintenance in Distributed-Authority Aviation Environments
Modern aviation maintenance operates within increasingly data-intensive technological environments, yet the operational integration of predictive maintenance into routine decision-making remains incon…
- arXiv 2025 · arXiv preprint
ROSflight 2.0: Lean ROS 2-Based Autopilot for Unmanned Aerial Vehicles
ROSflight is a lean, open-source autopilot ecosystem for unmanned aerial vehicles (UAVs). Designed by researchers for researchers, it is built to lower the barrier to entry to UAV research and acceler…
- arXiv 2025 · arXiv preprint
ROSplane 2.0: A Fixed-Wing Autopilot for Research
Unmanned aerial vehicle (UAV) research requires the integration of cutting-edge technology into existing autopilot frameworks.
- Semantic Scholar 2025 · Article (Applied Sciences)
Decision-Making Framework for Aviation Safety in Predictive Maintenance Strategies
The implementation of predictive maintenance (PM) in aviation presents unique challenges due to strict safety requirements, complex operational environments, and regulatory constraints.
- NASA NTRS 2024 · Technical Memorandum (TM)
Testing of Advanced Capabilities to Enable In-time Safety Management and Assurance for Future Flight Operations
In order to refine an initial Concept of Operations, explore Concepts of Use, and expose/validate requirements for future In-Time Aviation Safety Management Systems (IASMS), testing architectures were…
- arXiv 2024 · arXiv preprint
A Data-Driven Autopilot for Fixed-Wing Aircraft Based on Model Predictive Control
Autopilots for fixed-wing aircraft are typically designed based on linearized aerodynamic models consisting of stability and control derivatives obtained from wind-tunnel testing.
Browse the full corpus — academia portal ↗