As the Crow Flies vs by Road: How Much Longer Trips Really Are

Ask how far Tehran is from Mashhad and there are two honest answers. A crow would fly 743 km. A car on the fastest route drives 899 km. The gap between those two numbers is what this article measures.

We routed 116 pairs of cities in Europe, Iran, Russia, Turkey, Japan and Indonesia and divided each road distance by the straight-line distance. At the median, the road was 23.5% longer than the crow's line. Half of all pairs landed between 16.5% and 32.9% longer. The extremes come from water and mountains: Hiroshima to Matsuyama is 67 km as the crow flies and 193 km by road.

What "as the crow flies" means

Distance as the crow flies is the shortest path between two points over the surface of the Earth. It ignores roads, rivers, borders and terrain. On a sphere, that shortest path is an arc of a great circle, which is why the same number is also called the great-circle distance or the straight-line distance.

A flat map can make that line look curved. A long straight line drawn on a Web Mercator map is usually not the shortest route, for the reasons covered in our article on map projections. The distance itself only depends on the two coordinates, their latitude and longitude.

The road distance is what you drive. It follows the network, so it can never be shorter than the straight line, and in practice it is always longer.

The detour index

Transport researchers call the ratio between the two the circuity factor or detour index:

detour index = road distance / straight-line distance

An index of 1.00 means the road is perfectly straight. An index of 1.25 means you drive 25% further than the crow flies. The idea is old. Ballou, Rahardja and Sakai published circuity factors for a selection of countries in 2002, so that road travel distances could be estimated from coordinates alone.

What changes the index is geography. A motorway across a plain keeps it low. A mountain range, a bay or a strait without a bridge pushes it up, sometimes a lot.

How we measured it

On 29 September 2026 we picked 119 city pairs: 38 inside Germany, France and Italy, 20 more across Europe, 27 in Iran, 11 in Russia, 10 in Japan, 7 in Turkey and 6 in Indonesia. For each pair we:

1. Fixed one point per city, in the historic centre or near the main square.
2. Computed the great-circle distance with the haversine formula and a mean Earth radius of 6,371 km.
3. Requested the fastest car route from the MapAtlas Directions API with default settings and took its length and estimated driving time.
4. Divided road by straight line to get the detour index.

Three routes used a ferry or a car shuttle: Messina to Reggio Calabria across the Strait of Messina, London to Paris across the Channel, and Surabaya to Denpasar across the Bali Strait. We left them out of every statistic below, which leaves 116 pairs. The driving times are free-flow estimates from the router, without live traffic.

Results by region

France and Germany sit lowest among the European sets because their motorway networks radiate from large cities across mostly open country. Russia's median is lower still: the corridors we sampled, Moscow to Saint Petersburg, Nizhny Novgorod, Yaroslavl and Voronezh, run across flat land on roads that follow the direct line closely. Turkey and Japan sit highest, because their big cities are separated by mountains and inlets of the sea.

The samples are small per region, so read the medians as a description of these pairs, not of each country's whole network.

Straight line and road, pair by pair

A cross-section of the 116 pairs. Distances in km, driving time as estimated by the router:

Why some trips are so much longer

Ten of the 116 pairs have a detour index of 1.5 or more. Every one of them has water or mountains on the direct line.

• Sea straits and bays. Hiroshima and Matsuyama face each other across the Seto Inland Sea. The fastest road route runs east to Onomichi and crosses the islands on the Shimanami Kaido bridges, nearly three times the direct distance. Osaka to Kochi (1.57) goes west to Okayama to cross on the Seto-Ohashi Bridge. Istanbul to Bursa (1.69) drives around the Gulf of İzmit over the Osmangazi Bridge, and Copenhagen to Malmö (1.67) detours to the Øresund Bridge.
• Mountains. Krasnodar to Sochi (1.77) crosses the western Caucasus to the coast at Dzhubga and then follows the Black Sea shore. Trabzon to Erzurum (1.65) climbs over the Pontic Mountains, Isfahan to Ahvaz (1.61) and Shiraz to Bushehr (1.60) cross the Zagros, and Oslo to Bergen (1.57) crosses the Norwegian mountains between fjords.
• Coastlines. Athens to Thessaloniki (1.65) has the gulfs of the Aegean on the direct line, so the road bends west around them.

Short trips run higher too. Pairs under 150 km apart had a median index of 1.29, against 1.21 to 1.24 for longer pairs. City streets at both ends and the ramps onto a motorway add a similar number of kilometres to every trip, and over a short distance that overhead weighs more.

The straightest routes are the opposite case: flat land and a motorway built along the direct line. Moscow to Nizhny Novgorod (1.06), Milan to Bologna (1.07), Tehran to Qazvin (1.08) and Moscow to Tula (1.08) all stay within 8% of the crow's line.

From straight line to driving time

Distance is half the question. Across the 116 pairs, the median route covered 78 km of straight-line distance per hour of driving, at a median road speed of 95 km/h. So the crow's distance divided by 78 gives a rough first estimate of free-flow driving hours: 743 km from Tehran to Mashhad gives 9.5 hours, against 9 h 09 from the router.

Regions with mountains are slower in both ways at once: longer roads and lower speeds. The median straight-line progress was 91 km per hour in France and 66 km per hour in Japan.

A rule of thumb, and when it fails

If all you have is a straight-line distance, multiply it by 1.2. For our pairs that estimate came within 10% of the real road distance 83 times out of 116, and within 20% for 106 of them. A factor of 1.25, closer to the median, did worse at the 10% level (72 pairs), because the distribution has a long tail on the high side and many ordinary trips sit between 1.1 and 1.2.

The ten pairs outside 20% are exactly the ones with a sea strait or mountain range in the way. That matches what Boscoe, Henry and Zdeb found for more than 66,000 locations in the United States: straight-line and road distance to hospitals were very highly correlated (r² above 0.9), with the important exceptions near shorelines and other physical barriers.

So the practical rule:

• Use the straight line to rank or filter places, for example "stores within 20 km", when a few percent of error does not matter. It costs nothing to compute.
• Use a road distance when a person will drive the trip, a price depends on it, or the area has water or mountains. Request a route, or a distance matrix for many origins and destinations at once.
• Use travel time when the question is really "how long": a travel time map shows what is reachable in a given number of minutes, which a circle drawn on a map cannot.

Measure both yourself

For the straight line, open the distance calculator, click the two places and read the result in kilometres and miles. For the road distance, request a route. This is the request body we sent to the MapAtlas Directions API for each pair, here for Tehran to Mashhad:

{
  "locations": [
    { "lat": 35.6892, "lon": 51.3890 },
    { "lat": 36.2605, "lon": 59.6168 }
  ],
  "costing": "auto",
  "units": "kilometers",
  "directions_type": "none"
}

The response's trip.summary.length is the road distance in kilometres (898.7 here) and trip.summary.time the driving time in seconds. Divide the length by the straight-line distance (742.9 km) and you have the detour index for that trip, 1.21.

Method and limits

• Sample: 119 city pairs chosen to cover large cities, typical corridors and known barriers, routed on 29 September 2026. It is not a random sample of trips, so the medians describe these pairs.
• Points: one fixed city-centre coordinate per city. A different start address inside the same city changes short trips more than long ones.
• Road distance: the fastest car route with default settings. The shortest route would give slightly lower indices. Ferry and car shuttle routes (3 pairs) are excluded from all statistics.
• Straight line: great-circle distance on a sphere with a mean radius of 6,371 km. On the WGS84 ellipsoid the figure differs by well under 1% for these distances, which does not change any index at two decimals by more than 0.01.
• Time: free-flow estimates from the router, without live traffic, border waits or breaks.

To see how far the road network lets you get from one address in a set time, try the travel time map, and read what an isochrone is for how those areas are calculated.