On January 29, 2025, at 8:48 PM EST, a collision be-tween PSA flight 5342 (a CRJ-900) and Pat 25 (an Army UH-60L Blackhawk helicopter) represented the first fatal airline mishap in the United States in seven years (the previous being a single fatality associated with a shattered passenger window on Southwest flight 1380), and the first mass casualty event since Colgan Air flight 3407 in 2009. Two fatal crashes stretching sixteen years, a remarkable safety streak considering it covered nearly 140 million departures. In contrast, between 1993 and 2008, there were 18 fatal airline crashes across a similar number of departures.
The path to today’s safety record is clear in the data. In the 1970s, most fatal crashes stemmed from CFIT or microburst-induced windshear. By the 1980s, terrorism and violence overshadowed other causes, with only two accidents outside those categories—types now largely prevented by EGPWS and takeoff configuration alerts. The 1990s marked a turning point: TCAS, EGPWS, and configuration warning systems virtually eliminated traditional accident types, while the 1994 crash of USAir 1016 led to windshear detection networks that ended weather-related losses. Since the 1970s, terrorism has led airline fatalities by far, with a spooky 666 lives lost in seven crashes, plus another 2,977 victims during 9/11. The shock of the 2001 event resulted in massive changes to airport security that have effectively eliminated airline terrorist attacks.
A series of midair collisions between 1969 and 1978 led to the development of TCAS (Traffic Alert and Collision Avoidance System). But this was not aviation’s first encounter with fatal collisions—the 1956 crash between TWA Flight 2 and United Airlines Flight 718 was the deadliest U.S. air disaster of its time. Congressional hearings and public outcry resulted in the formation of the FAA (in 1958) as well as funding for a national radar-based air traffic control system. Prior to this, aerial collisions were prevented mainly through a combination of ‘see-and-avoid’ and ‘big skies.’ In 1960, in response to a second fatal collision between TWA and United Airlines, the integration of rudimentary transponders into the national aerospace system was mandated by the FAA. The 1978 collision of PSA 182 and a Cessna 172 over San Diego resulted in Mode C transponder requirements extending also to GA aircraft while operating in Class B airspace.
Before the Potomac collision earlier this year, the last U.S. airline midair fatality occurred in 1986, when a Piper Archer penetrated Class B (then TCA) airspace without clearance and struck an Aeromexico DC-9 over Cerritos, California. All 67 aboard the DC-9, three in the Piper, and 15 people on the ground were killed. The outcome was sweeping reforms: TCAS became mandatory on all U.S. Part 121 aircraft, and Mode C transponders were required for all aircraft within the lateral boundaries of Class B airspace, regardless of altitude. Though this crash effectively ushered in a welcome forty-year hiatus in aerial collisions involving the airlines, corporate and GA aircraft still occasionally tangled. In response, ATC software was introduced to provide aural alerts warning controllers of unsafe separation.
Tragic Return to Reality
The happy drought in U.S. airliner fatalities was bound to eventually come to an end. The first observation from the Potomac crash is an old one. “See-and-avoid” is unreliable and cannot solely be counted on for collision avoidance. The NTSB has not yet released its final report, but it will likely assign some responsibility to the pilots for failing to visually identify one another. While TCAS is officially designated as ‘a last-resort safety barrier,’ midair collisions persisted until electronic alerting systems became mandatory. While pilots can and do ‘save the day’ by visually spotting and avoiding intruders—and it is right that we are reminded of this duty—we will forever remain an imperfect safeguard against midair collisions.
Air traffic control’s primary responsibility is to facilitate aircraft separation and prevent collisions. In busy airspace, however, controller workload can be overwhelming, and certain responsibilities may be delegated. For example, when a controller in terminal airspace asks a pilot if they have another aircraft in sight, they are attempting to transfer separation responsibility: if the pilot responds ‘no contact,’ the controller retains it; if the pilot responds ‘in sight,’ the pilot assumes it. There is an inherent danger to this. In busy terminal areas, there may be several aircraft in roughly the same quadrant of the sky, all converging on a similar point. How can a pilot be certain they have the correct aircraft in sight? Controllers often refer to distant traffic, which is not yet visible. It is easy for pilots to mistake a closer, more visible aircraft as their target. This happened to me years ago on approach to Missoula Airport, Montana. Tower instructed me to track an aircraft on final, and I visually identified one on the bearing provided and turned base to follow. Concurrently, I noticed another aircraft depicted on TCAS further out—the actual traffic I was supposed to trail. Despite experience and vigilance, I had misidentified the target. Without TCAS, we likely would have remained on a collision course. The airport was to my right, and I doubt I would have looked back to the left in time to see the correct aircraft.
This see-and-avoid dynamic was intermixed with others the evening of the crash. The Army Blackhawk was being used for a check ride, with the check pilot sitting in the left seat and the evaluated captain (who outranked him) sitting in the right (in helicopters, the PIC occupies the right seat). The voice recording transcript of the flight depicts a deteriorating evaluation. An abnormal procedure was performed inappropriately by the captain, followed by a missed approach to a landing site due to an unstable profile (the instructor pilot called for the go around). The instructor was forced again to intervene when the captain flew an improper go around profile. During the second approach attempt she stated, “I’m not comfortable landing there.” The check pilot took the controls and demonstrated the landing and asked the captain if she would like another try, which she declined.
In the vocabulary of checkrides, the sequence was clearly unsatisfactory. The implications likely weighed on the check pilot. The captain not only substantially outranked him, but she held a coveted spot as a White House Military Social Aide, where she rubbed shoulders with dignitaries, celebrities, and the President of the United States. Paired to this pedigree was a somewhat muddled history in the Blackhawk. A battalion standardization pilot who had conducted an instrument evaluation with her three years prior had “found her performance well below average. He described her as one of the bottom three pilots he had trained” (NTSB Human Performance Factual Report). All this likely resulted in an uncomfortable situation for the instructor pilot, which may partially explain his decision-making during the fatal sequence of events that followed.
As Pat 25 received DCA tower approval to transit via Helicopter Route 4 (more on this later) the controller asked him to report the PSA regional jet in sight, 7 miles ahead. At this point, there were several aircraft lined up for Runway 01, including PSA 5342 (the circle to Runway 33 is based on initial alignment with Runway 01). The initial communication from ATC to Pat 25 was broken, indicating that the crew may not have understood the regional jet was circling to Runway 33. They may have assumed it was among the long line of arrivals for Runway 01—which, critically, does not conflict with Route 4. The instructor transmitted that he had the traffic in sight, but there was no way for him to know which of the many landing lights belonged to the referenced aircraft. Adding complexity, the checkride required the use of night vision goggles (NVG). While NVG units improve visibility in low-light conditions, they also substantially restrict peripheral vision. This would prove critical as the regional jet would end up on a collision course to the far left of the Blackhawk.
Not long after this, the tower controller apparently received an aural conflict alert between the helicopter and the regional jet. He queried Pat 25 if they had the CRJ in sight, but he did not provide either range or bearing to the intruder. The helicopter and CRJ were separated by less than a mile and rapidly converging, critical information that was omitted. The instructor pilot stated he had the CRJ in sight—likely assuming, incorrectly, that the reference was to an aircraft on final to Runway 01. The controller instructed the crew to “pass behind the CRJ,” but a concurrent radio transmission blocked this instruction. Around this same moment, the CRJ crew received a TCAS traffic alert (an aural “traffic, traffic”). With less than a mile separation (and at low altitude), TCAS would normally issue a climbing Resolution Advisory, which the CRJ crew was extensively trained to perform. However, to reduce hazards in congested airspace, TCAS does not provide Resolution Advisories below 1,000 feet AGL. Instead, the CRJ received only the lower-level Traffic Advisory, which comes with an explicit caveat: crews should not maneuver solely based on this alert. Instead, they should maintain their course and attempt to visually identify the traffic. Although the Blackhawk’s anticollision lights were on, its relative angle to the CRJ may have caused it to key into the background of city lights. On a collision course, an object has no relative motion—only a gradual increase in size—something the human eye is especially poor at detecting at night. Although the CRJ crew did not verbalize searching for the traffic, the aural alert makes it highly likely they were. Unfortunately, a combination of high workload and physiological limits was stacked against them.
Helicopter Route 4
With heavy civil and military traffic around Washington D.C., the FAA approved dedicated helicopter routes near DCA to ease controller workload. These routes left little margin, with tolerances ranging from tight to nonexistent. Pat 25 joined Route 1 at Cabin John, which follows the Potomac River with step-down fixes toward DCA. This design is optimized for south-flow operations: fixed-wing arrivals to Runway 19 fly above the river with altitude floors, while helicopters remain below with altitude ceilings, the difference between the two assuring vertical separation.
On the night of the accident, winds gusted 14 to 23 knots from the northwest, putting the field in a relatively simple north-flow configuration. Most traffic was using Runway 01, shared by both arrivals and departures. With the bank of traffic growing, controllers opened Runway 33 for some landings. At just 5,204 feet, it is short by airline standards. Fewer than 5% of landings occur on 15/33, and when used, it is almost always limited to lighter regional aircraft. Using intersecting runways boosts flow rates slightly, allowing more takeoffs and landings per hour—but at the cost of significantly higher workload for both controllers and pilots. Because of Runway 33’s short length, controllers typically ask whether pilots can accept it rather than simply assigning it—as occurred this night. After some initial hesitation (followed by verification of sufficient landing performance), the PSA crew accepted the clearance. This detail later weighed heavily on the first officer’s mother, who told a pilot union representative at an NTSB briefing that she agonized over the crew’s decision not to refuse. In practice, however, for DCA-based pilots, accepting the shorter runway was routine.
To facilitate southbound helicopter traffic, Route 1 transitions to Route 4 at a fork in the Potomac. Route 4 is designed to provide separation for traffic landing on Runway 01, but it does not provide acceptable margins to Runway 33. Even when flown tightly along the eastern bank, Route 4 passes less than 100 feet below a standard glidepath to Runway 33. The best-case scenario is a near miss. Compounding the risk, Route 4 lacks defined lateral boundaries. Just a hundred meters west (the approximate flight path of Pat 25 that night) is a bullseye to the approach path for Runway 33. Collision becomes a matter of timing.
Simultaneous use of Runway 33 and Route 4 depends on helicopter crews visually acquiring and avoiding Runway 33 traffic. Task-saturated airliners performing a low-altitude circle to 33 will not be able to monitor the approach while also devoting substantial time to searching for traffic. While collision avoidance is officially every pilot’s responsibility, its limits are well known. ATC itself was created to compensate for the shortcomings of “see-and-avoid,” and TCAS was later introduced to cover the gaps left by both pilots and controllers in reliably preventing midair collisions. Yet Helicopter Route 4—intended to provide safe transit near DCA during 01/19 operations—remained available even when aircraft were circling to land Runway 33. See-and-avoid as a safeguard had already been impeached—long shown to reduce but not eliminate fatalities. As the primary barrier against calamity, it was only a matter of time.
There were other complications. It is doubtful that the architects of Route 4 considered helicopter training events under NVG, and the peripheral limits such devices produce on collision avoidance. Likewise, operating a helicopter at 200 feet, on NVG, along the banks of the Potomac—with cranes and buildings speckling the shore—is itself task-intensive. At such a low altitude, the flying pilot must focus almost entirely on precise margins. That shifts radio calls, decision-making, and visual scanning onto the monitoring pilot. With the flying pilot low on bandwidth, there is no room for misunderstanding or error. The error chain is complete; survival or catastrophe left to chance.
Route 4 has been killed by the FAA, which represents a particular solution and not a general one. See-and-avoid is relied on elsewhere, such as Orange County Airport in Santa Ana, California. With extremely close parallel runways and a constant mix of airline, corporate, and GA traffic, it is common to receive a traffic alert on short final as parallel base-leg aircraft are briefly on a collision course. Harrison Ford famously landed on a taxiway there, good for an eye-rolling laugh if you have never experienced the tight confines of the SNA traffic pattern, nor the taxiway-like width of Runway 20L. On my last two approaches to 20R into SNA, I have received low altitude TCAS alerts both times. Straight-in during daylight, it is easy to spot the offender and verify their last-minute turn to final. It is disconcerting, but there is no other alternative. Aviation demands precision, yet there is no getting around risk. Technology has made aviation safer, but it is not infallible. Every system—electronic, mechanical, and human—has limits, and every environment comes with unique challenges. Complacency erodes diligence, yet every flight demands respect. Complex airports, conditions, or equipment require special preparation. The Potomac collision is a stark reminder that while technology and regulation have driven aviation to unprecedented levels, the final defense still rests on human judgment—and when that judgment leans too heavily on see-and-avoid, history shows it is only a matter of time before blood is spilled again.
