These days, on larger/newer aircraft at least, GPS navigation is ubiquitous. So VORs and Non-directional Beacons (NDBs) are used less. Wasn't the reason for airways because they were the connection points between VORs and NDBs?

It seems like a pilot could easily "fly direct" from one airport to another without using any intermediary checkpoints. And yet I've been told on several occasions that most pilots use airways.

Why do pilots still use airways instead of flying direct?

  • 8
    $\begingroup$ I can't seem to find any references at the moment, but I distinctly remember an Air Crash Investigation episode that stated that flying direct was the norm in the early days of aviation; until a spate of accidents forced the FAA to establish airways to control this. $\endgroup$ Jan 5 '15 at 23:39
  • $\begingroup$ Interesting, if you can find it @DannyBeckett, I'd be curious to see it. $\endgroup$
    – Jay Carr
    Jan 6 '15 at 0:41
  • 1
    $\begingroup$ @DannyBeckett That may well be true but the original reasons for establishing airways might no longer apply. For example, if it was just to make navigation easier, major airlines would probably turn around and say, "Thanks but we have GPS and computers and stuff. We don't need that help." $\endgroup$ Jan 6 '15 at 8:56
  • 12
    $\begingroup$ On a flight from JAC-MCO (~1700nm) in a PC-12 (NBAA IFR Range 1600nm), I filed a route on the backside of a low which was hovering over the direct route. Since ZLC is so sparse, every new controller gave us "Cleared direct MCO," and I kept telling them we want "as filed." We made MCO with plenty of fuel to spare, and it was the longest endurance flight we ever did (yes, it had an onboard potty). $\endgroup$
    – rbp
    Jan 6 '15 at 15:22
  • 1
    $\begingroup$ @Jay Carr...I think the episode in question dealt with a collision between two airliners over the Grand Canyon in the '50s(?). Both flights deviated to give passengers a view.. $\endgroup$
    – DJohnM
    Jan 7 '15 at 2:29

Airways simply allow for better management of traffic.
Imagine for a moment that everyone had an off-road capable car, if all the drivers were going "GPS Direct" to their destination how would drivers ensure separation? How would you avoid hitting other cars if there were no roads?

Airways are the aviation solution to this problem: Defined routes between navaids (VORs, NDBs, and the like). Airways can also be defined using "fixes" defined by GPS coordinates which can provide something closer to "GPS Direct" routing while still providing a defined route.

Airways also provide a bunch of other benefits, including:

Traffic Flow Management

Flightplans in Europe are validated by the CFMU (Central Flow Management Unit) and checked for criteria such as direction of airways (which can change during the day, e.g. one airway being for eastbound traffic only between 1000 UTC and 1400 UTC), correct flight level ranges and airways correctly connecting waypoints, such as fixes, VORs or NDBs. The bottleneck in all flight operations is understanding airport capacity (how many operations/hour) and sector capacity (how many aircraft handled per sector/per controller). The goal is sometimes to distribute traffic between two sectors, although having the same destination. Below you will find two routings to Düsseldorf - EDDL, one departing from Leipzig - EDDP and one from Berlin-Tegel - EDDT (click on links to see visual representation):



The routing from EDDP to EDDL is going via WRB - TINSA - ADEMI - INBAX - DOMUX, which is within the boundaries of a sector called "Paderborn High" or PADH.

The routing from EDDT to EDDL is going via PIROL - DENOT - HMM - XAMOD, which is within the boundaries of a sector called "Hamm High/Medium" or HMMH/HMMM.

During peak hours, both sectors are separate and deal with traffic only within their sector, which uses the full capacity of each controller handling the sector. Outside peak hours, the sectors can be combined and worked by one controller. This allows a granular usage of controller resources and airspace.

Separation between Traffic

From the example above we have seen that there is 2 routing endpoints at Düsseldorf - EDDL for traffic arriving from the North-East and South-East, XAMOD and DOMUX. Both waypoints or fixes are respective STAR or transition entry-points for arrivals into Düsseldorf. By using a standardized routing to and from airports, we can now expect traffic to always arrive via these two points from the east, unless other coordination has been achieved by the ATC units involved. Looking at the reverse routing for both airports, we will see these routes:



MEVEL and NUDGO are two of the SID exit waypoints out of Düsseldorf, MEVEL being 10nm north of XAMOD and NUDGO being 26nm east of DOMUX, however the relevant fix here is ELBAL, which is one of the fixes on the SID23L from Düsseldorf to NUDGO, which is 16nm south of NUDGO. With something simple as using standardized entry and exit fixes to and from an airport, we have managed to maintain a traffic flow with ensured separation between departing and arriving traffic, something which would not be possible with GPS direct routing or would require constant ATC vectors.

Handoffs from ATC Unit to ATC Unit

The above examples show different routings and sometimes different sectors handling traffic on these routes. How is traffic handed off to other controllers and where? A handoff consists of moving the radar track of the aircraft from one controller to another and thereafter instructing the aircraft to change the frequency to the next sector controller. Handoffs are initiated at sector boundaries, at specific fixes, which have been agreed on between different ACC (Area Control Centers) or sometimes even within one ACC between single sectors. Since we are already familiar with Düsseldorf airspace, let's use the below routing from München to Düsseldorf.


From the graphical represenation you will see that traffic from EDDM to EDDL will pass the waypoint ARPEG, which is close to the destination airport. Before reaching the waypoint ARPEG, aircraft will be under control of either the Hersfeld (HEF) or Gedern (GED) sector (page 5) and need to descend towards the destination airport. The next sector for arriving traffic into Düsseldorf via ARPEG would be Paderborn High Sector (PADH). The agreements between HEF/GED and PADH are that arriving traffic to Düsseldorf via ARPEG is to be expected at FL240 level at ARPEG and released after handoff for further descent through PADH, even if the aircraft has not yet crossed the sector boundary and is not in PADH airspace. These coordination waypoints and handoff procedures are documented in Letters of Agreement between or within an ACC.


When weather (wrong-way lows, strong jetstreams, storms, etc) affects the Direct route, flight dispatchers will pro-actively suggest routes around it. It's not unusual for NYC-SFO traffic to be routed over Canada or over Kentucky when the Midwest is storming.

Exceptions - Free Route Airspace

Where airport and air traffic density allows, the use of airways is not mandatory or in some cases, not even expected or provisioned for. One example is the Free Route Airspace overhead Scandinavia, where a portion of airspace can be freely used between defined entry and exit points. Similar systems are also being discussed by NATS UK and the FAA NextGen Air Transportation System.

For more information on airways and their use, see also the related question: Is there a difference between how commercial jets and GA aircraft use airways?

Some material courtesy of VATSIM or VATSIM Germany. While only for simulation use, the material used is as accurate and close to the real sectorisation or procedures as possible, so should be sufficient to explain the concepts presented here.

  • 7
    $\begingroup$ Excellent car analogy $\endgroup$
    – Jon Story
    Jan 5 '15 at 23:51
  • 12
    $\begingroup$ Roads aren't there for traffic control; they're there because cars are inefficient and (even somewhat more than usually) unsafe off-road, with or without any other cars around. This analogy seems highly dubious. $\endgroup$ Jan 6 '15 at 3:28
  • 4
    $\begingroup$ @NathanTuggy I have amended my analogy, I had the assumption in mind that all cars were off-road capable, same as all aircraft are flight-capable. Otherwise the analogy wouldn't make sense, as you have pointed out. $\endgroup$ Jan 6 '15 at 7:32
  • 4
    $\begingroup$ The Say Again series on AvWeb discusses using the routes many times over. My understanding from it is that ad-hoc routes are not big problem in radar environment, but nothing but departure and destination airports is. At the very least the flight plan should have departure point (wehere SIDs end), initial approach fix (where STARs start) to destination and one point after each sector transition so the controller will have a known reference point for your track and point to hand you over at. $\endgroup$
    – Jan Hudec
    Jan 6 '15 at 11:31
  • 2
    $\begingroup$ @JanHudec The Say Again series are dominantly FAA-based, so I can't comment. In Europe, you will rarely see non-CFMU validated flightplans due to the complexity and density of the airspace. $\endgroup$ Jan 6 '15 at 12:45

A simplistic explanation is, pilots in controlled airspace fly along airways because that's where the air traffic controllers want them to fly.

The use of airways has to do with the way air traffic controllers keep track of all the flights in the sector of airspace they control (they need a way to organize the traffic so that they have a good mental image of it and can guarantee that aircraft do not pass too close to each other), as well as the fact that it is easier to communicate a path that goes between published waypoints with three- or five-letter names than a path that goes between points described by their numeric latitudes and longitudes.

The free flight concept proposes to allow these same aircraft to fly routes that do not generally follow airways, but it requires big changes in the way air traffic management is done.

(This answer is based primarily on recollection of the work I've done on the development and analysis of air traffic management systems, mostly related to free flight, since 2001.)

  • $\begingroup$ "because that's where the air traffic controllers want them to fly" -- I think this is not true historically. Many airways were developed because of terrain, radio coverage, and city-pair traffic. $\endgroup$
    – rbp
    Jan 6 '15 at 18:14
  • $\begingroup$ @rbp I did say "simplistic". The point of the OP, however, which I think was valid, was that the available choices of flight paths are much more limited (most of the time) than current navigation technology could support. The reasons I usually hear cited for this basically come down to safety, specifically, maintaining separation between multiple aircraft, which is a very different problem than the problem of navigating an individual aircraft. $\endgroup$
    – David K
    Jan 7 '15 at 2:55
  • $\begingroup$ @rbp P.S.--Loved the story about your JAC-MCO flight. $\endgroup$
    – David K
    Jan 7 '15 at 2:59
  • $\begingroup$ fair enough. Glad you liked the story $\endgroup$
    – rbp
    Jan 7 '15 at 4:44

Airways are safer, because you know where other pilots will be flying in order to avoid them. The rule of using an odd flight level for flying east, and an even flight level for flying west will also help to keep traffic separated.

Safe cruising altitudes are also set based on the airways. There is an altitude per waypoint so a normal climb will let you get over the next hill, and per section between waypoints so you know how high that hill actually is.

Without the airways system maintaining proper ground clearance would be much more work intensive. With the fixes all the pilot needs to do is keep a list with fixes and MEA/MCA (safe crossing altitudes) and follow the needle and cross them off as he reaches them and climb if necessary.

  • $\begingroup$ In the absence of separation between eastbound and westbound flights, would collisions be more likely between eastbound and westbound flights, or between flights going in roughly the same direction? I would think the dangerous scenario from a collision standpoint would be two aircraft flying almost parallel such that neither one ever entered the field of view of the other. $\endgroup$
    – supercat
    May 20 '15 at 23:14

Just to add a little perspective, route-free airspace has been in consideration for a while and is already implemented in certain countries. The route-free or direct-routing concept makes use of the currently underutilized satellite systems on aircraft (In the routing context) and intelligent software to execute more efficient routing of aircraft (GPS-based). Direct-routing is also a part of the the NATS UK and the FAA's NextGen air traffic management strategy.


Speaking as a GA pilot flying light airplanes in the US system:

On flights in lower use airspace, at lower altitudes, and with the appropriate ATC controls in place (agreements allowing controllers to do that across into someone else's airspace) it's possible to get a filed GPS direct route approved. I've done so on trips in the 200-300 mile range before.

On longer trips, you don't necessarily have to use the airways. It's often possible to get waypoints at the ingress/egress of airspace, which you can go direct between (off airway).


It's like a lot of systems that evolve. If we started today, we might do it differently.

The analogy with cars is a poor one because, for the most part, there are a very small number of aircraft in a very large amount of airspace and it would be hard to justify the use of airways. So starting from the standpoint that every aircraft accurately knows:

a) Its location, heading, velocity vectors, fuel data, flight constraints and destination etc;

b) The same information for all other aircraft in its vicinity (this could cover quite a large volume in practice);

c) The terrain and weather conditions etc.

Then it becomes a really interesting combinational problem to solve in the context of an autonomous distributed computer system; i.e. no ATC - all handled by cooperating in-flight systems.

Clearly there are significant points where paths must cross because of the geographical positions of airports, and of course the dynamic scheduling of arrivals and departures at airports; not to mention the handling of unexpected events.

For a computer scientist this is all fascinating and a meaty challenge, but I suspect that it would be very difficult to develop a system of autonomous control based on this approach to the point that it could be trusted!

  • $\begingroup$ What statistical data do you base the assumption on that there is a small number of aircraft in a very large amount of airspace? Especially around busy arrival airspaces, this is quite not true. $\endgroup$ Jan 7 '15 at 12:39
  • 1
    $\begingroup$ The question was about the need for airways in general, not arrival airspaces. Even taking the busiest time for the busiest country in the world, aircraft (with a reasonable separation) currently occupy only a tiny fraction of the available space. Even close to a busy airport with up 200 flights/hour, a basic cubic law indicates that any congestion need only be restricted to a very short distance around it. $\endgroup$ Jan 7 '15 at 13:53
  • 1
    $\begingroup$ Not sure I agree with "a" completely. A light GA (without GPS) aircraft knows it's general location, but to what numerical value? Don't properly adjust your altimeter for change in weather and maybe you're off by a few ten's of feet. There ARE still aircraft out there flown by pilots without GPS, without even a radio. I also totally disagree with "b". My father, flying his old Piper, has no idea of all other aircraft in his vicinity. He knows approximately the locations of those he can see visually, but that is it. $\endgroup$
    – CGCampbell
    Jan 7 '15 at 14:10
  • $\begingroup$ I was making the assumption that at some point it will be automatic and mandatory for all aircraft, much as it will be for cars on highways in the near future. $\endgroup$ Jan 7 '15 at 23:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.