Top end road bikes can be broadly split into two categories, aero bikes and lightweight climbing bikes. Aero bikes such as the Giant Propel, Merida Reacto and Cervelo S5 are often heavier than their climbing counterparts.

In addition, they can be less stiff and less comfortable. Considering this we wanted to see if an aerobike is actually faster and if so, by how much? Is the slight increase in weight worth it?

>>> Lighter vs aero, which is better?

To do this we headed down to Herne Hill Velodrome in South London, to put two bikes to the test. The bikes we chose for this experiment were the wind cheating Cervelo S5 as the aero bike and the svelte but stiff Canyon Ultimate CF SLX.

The Cervelo S5 is often considered to be amongst the most aerodynamic road bikes, featuring aero handlebars, Hed Jet 6 wheels, aero tube profiles and Shimano Dura Ace Di2 groupset. It is currently ridden by British Cycling, One Pro Cycling and MTN-Qhubeka. The weight of our S5 is 7.25kg without pedals.

>>> The lightest production bike in the world (video)

The Canyon Ultimate CF SLX is the same type of frame as ridden by Movistar and Katusha. It is a race proven winner, that Nairo Quintana rode to victory in the 2014 Giro d’Italia. Our test bike is fitted with Campagnolo Super Record and Mavic Kysrium R-SYS wheels. The Ultimate is roughly a kilogram lighter than the S5.

>>> Buyer’s guide to road bike wheels

Bikes + Hutch

The Canyon, left and the Cervelo, right

Our Experiment

Each bike was ridden for 10 minutes at 200W and 10 minutes at 300W. The reason for this is that we felt that 200W represents a power output sustainable by a decent club or sportive rider and that 300W is typical of a time trialist or racer. From this we wanted to establish if it is still worth getting an aerobike if you don’t ride at race pace.

Both bikes were ridden by Dr Hutch and both bikes were set up with the same position, so that the rider remained constant and that only the bike changed. Hutch rode exclusively on the hoods and maintained the same position throughout all the test runs.

It may be of benefit to Mark Cavendish in a 60kph sprint, but what about us mortals? Is there a measurable aero benefit at 200W and is the benefit greater at higher wattages? We chose Herne Hill because for all the marketing claims and wind tunnel data that exist, this isn’t the real world. You don’t ride your bike in a wind tunnel, you ride outside and Herne Hill represents the almost as much of a controlled outside environment you can get.

>>> We tested an altitude tent. Click here for the results

Herne Hill Velodrome - Aerobike Vs Road.00_03_23_01.Still004

Dr Hutch calmly sitting at 300W on the S5

The Results

Before I get onto the numbers, I need to point some things out. Firstly we are aware that this is not the most scientific experiment in the history of aerodynamic research and there are variables that are out of our control. Things to note are that it was an unusually still day at Herne Hill, the wind was low and consistent throughout the runs we did on both bikes.

Dr Hutch, a veteran of both the track and time trialling was an excellent human guinea pig, as his pacing was very smooth and ensured metronomic power output on each of the runs. Each run began with a rolling start, to ensure that Hutch was already up to the required power output.

For power measurement we used Garmin Vector pedals and a Garmin Edge 1000 head unit. The reason for this is that we wanted to use the same power meter for consistency and that Vectors are easily swappable between bikes. To ensure accuracy, we used a torque wrench to tighten the pedals to 25 lbs/ft, where upon they were calibrated using the Garmin Edge 1000.

The Canyon Ultimate CF SLX

  • At 200W the Canyon travelled 5150m at an average speed of 30.9kph
  • At 300W the Canyon travelled 6055m at an average speed of 36.3kph

The Cervelo S5

  • At 200W the Cervelo travelled 5425m at an average speed of 32.6kph
  • At 300W the Cervelo travelled 6490m at an average speed of 38.9kph

This result is significant, as even after just 10 minutes at 300W the Cervelo S5 was 435m ahead of the Canyon (nearly half a kilometre!). If we extrapolate this to 20 minutes or an hour the difference could be massive and hugely significant in a race situation such as a breakaway.

But for mortals just hoping to achieve a personal best on their favourite sportive, the potential savings appear huge too. After just 10 minutes at 200W the Cervelo S5 was 275m ahead. Over four hours this could 6.6km if you extrapolate.

>>> Power meters, everything you need to know

Before we all go out and buy aero bikes it is worth considering a few things. Firstly, some aero bikes may not be as efficient as others, and deep wheels on a lightweight bike could make a significant aero improvement. The Cervelo S5 is also substantially less compliant and comfortable than the Canyon too. We would also expect the Canyon to be quicker up hill as it is roughly 1kg lighter, so an aero bike may lose some ground here. To find out if an aero bike is demonstrably slower up hill, we will aim to do a further experiment in the future…


  • This difference may just be the results of the 10 cm high rimms and just a few spokes in the hed wheelset and not the frame. Would like to see the test with the same wheelset and tires

  • Marco

    I agree, totally unfair. the wheels have likely a much stronger aerodynamic impact than the frame. Even then, there still would be too many variables like groupset, riding position, etc.)

  • Rick Harker

    It is a feasible real world test. I’ve seen many data points about a specific aero bikes speed benefits over another brand but always in a wind tunnel and without a rider. I’ve always questioned how much difference with a rider and what percentage of drag is the rider on the bike.
    Some of us can only afford one bike and this test will help make an informed decision. Along with that though is comments relating to wheels and how much difference they can make and that should be aggregated into the decision making too.

  • James Gill

    there are just too many variables here. Different gears, drive trains. the only constants are the rider, location and pedals.

    The only thing it tells us is that one BIKE is faster than the other. No why.

    “We’ve designed an experiment”
    ? really ?? using what crayons and a wall??

    This borders on pseudo science with regard to scientific method and accuracy!!

  • Bob

    I cant see anywhere where it says anything about those tiny bubbles on an Aero bike smoothing out the airflow, much like the dimples on a golf ball I suspect, but then again I couldn’t be bothered to read it

  • Brendan

    yeah guys . . . the wheels. bit of a wasted test. the question you posed was about the bike/frame. not about the wheels. how much of this difference is because of the wheelset? in which case maybe we need wheels for speed and wheels for gravity and the frame may not be that important.

  • ESTrainSmart

    I think @adrianoconnor:disqus is looking for a quantifiable way to distinguish the individual beneficial effects of the wheel versus the frame. Based on my several search attempts, I’ve only found an abundance of subjective reports with unclear experimental protocols. I find it interesting that you’ve seen people perform this test hundreds of times because this is the first time I’ve seen a test with this experimental design. Until now, the tests I’ve seen only state time or watts saved over a fixed distance- all performed in controlled lab conditions, not outdoors like this one. Can you please share a couple links to the tests similar to this one? I’d like to see how the other test results compare to this one. Thanks.

  • Alex Simmons

    The article was not laterally powerful nor aerodynamically compliant.

  • djconnel

    On the introduction you state the bike is only 20% of the total drag. On a closed course with light wind the speed is roughly proportional to the cubed root of power. Thus a perfect aerodynamic bike would increase speed by at most the cubed root of 1.25: a 7.7% advantage. I thus think either the winds increased for the canyon rides, the rider position was different, and/or the vector pedals didn’t get a proper settling in time after remounting.

  • Jack Mott

    People have done this test 100s of times. Yes the frame alone makes a difference.

  • Jack Mott

    “The Cervelo S5 is also substantially less compliant and comfortable than the Canyon too.” Test this, use the same wheels, tires, tubes and pressures on both bikes, and blind the rider as to what bike they are on. Will make for a fascinating article.

  • Conscience_of_a_conservative

    No benefit if you’re the middle person in a pace-line, the extrapolations border on silly as most riders can’t keep up 300 watts for an extended period of time, let alone a full ride. Also for those parts of the ride that are below 20 mph such as on climbs again , no benefit. This stuff is more about selling the idea of being fast, than actually being fast.

  • David Mulligan

    I was just thinking the same thing.

  • adrianoconnor

    Please re-do this with same wheels on both bikes! Would love to know if frame alone made any difference… Been wishing somebody would do this test for ages — certainly beats the wind tunnel data manufacturers go on about in their marketing gumph. Would also be interesting to see power difference between the different bikes riding in pairs at a set speed and swapping on/off the front. Anyway, looking forward to the hill climb test too if/when you do the follow up…

    I need all of this data to justify a Canyon Aeroad… 🙂

  • 65juicer

    Was that a S5 that had the front end break on Yanto Barker at Pearl Izumi Series in Bath `?

  • I just can’t see that this test tells you whether the aero frame is quicker or not as two variables (frame + wheels) have been compared simultaneously. Surely the fact that an aero frame plus aero wheels is quicker than a non-aero frame and non-aero wheels on a flat course on a still day has been well known for years.

  • Romain Mousset

    You should have used a true power meter ( garmin pedals sucks ) and same wheels tubs on both bike

  • Chris Coulter

    I agree. Especially given that the Mavic R-SYS are widely regarded to be about as aerodynamically optimized as a refrigerator.

  • jship

    surely would have been better to have the same wheels on each bike then you can better test the aerodynamics of the frame?

  • Mark Higham

    Could you please do this test in the next week I’ve just had my insurance payout from getting knocked off a month ago and I’m considering both these bikes 🙂