Cycling is a beautiful sport. Weather, traffic, and terrain are part of every ride, but we believe cycling culture has normalized danger while the industry sells speed before safety.
The myth of the upstroke
For decades cyclists have been told to "pedal in circles", with clipless allowing you to use your "full" pedal stroke. But a more effective pull is not necessarily a more efficient way to ride. During steady cycling, deliberately pulling up reduced efficiency in two experiments. [1] [2]
Korff et al. (2007) tested eight male cyclists at 200 watts and 90 rpm, finding that actively pulling increased force effectiveness but lowered gross efficiency. [1]
More pull. Less efficiency.
Korff et al. (2007)Gross efficiency (%)
Mechanical output relative to metabolic energy used.
- Other techniques*
- 20.2%
- Deliberate pull-up
- 19.0%
*Preferred, circling, and push-emphasis techniques pooled. Gross efficiency: 20.2% ± 0.6 vs. 19.0% ± 0.7 (mean ± SD). This was a steady-state test, not a Bythlon comparison. [1]
What do other upstroke studies show?
Mornieux et al. (2008): coached pulling reduced net mechanical efficiency by 9% in eight elite cyclists and 3.3% in seven non-cyclists, relative to normal clipless pedalling at 60% of maximal aerobic power and 90 rpm. These are relative reductions, not percentage-point changes; net and gross efficiency are different measures. [2]
Edwards et al. (2009) found that a more even torque pattern was associated with lower efficiency in 11 trained cyclists at some workloads. The association does not establish that changing technique caused the difference. [3]
Pulling can produce useful force. Mornieux et al. (2010) found improved upstroke effectiveness and changed muscle coordination in nine elite cyclists at 320 W. Their proposed oxygen-cost penalty was not a measured efficiency result; they also raised the possibility of benefits from training the coordination. [4]
Theurel et al. (2012) tested two 45-minute rides with preferred technique or visual force feedback. Efficiency was lower with feedback during the first 15 minutes, but did not significantly differ thereafter. The feedback condition also produced a smaller decline in maximal power after the ride, suggesting a possible fatigue benefit. [5]
Duc et al. (2019) reviewed technique and training studies. Better pulling mechanics had not consistently improved gross efficiency or performance; the authors called for more research on long-term adaptation and different riding conditions. [6]
Elmer & Martin (2021) studied a champion cyclist with seven years of single-leg cycling experience. Adding a counterweight to reduce the need to pull up lowered metabolic energy use and raised gross efficiency. This was one rider in a specialised setup, not a direct test of ordinary two-legged cycling. [7]
García-López et al. (2016) found that professionals had less resistive torque during the upstroke than elite and club cyclists. That study measured pedalling mechanics, not metabolic efficiency; it does not show that professionals simply rested through the recovery phase. [8]
The downstroke remains central to the story. Coyle et al. (1991) found that elite-national cyclists generated their higher performance power primarily by applying larger forces during the downstroke than good-state cyclists. [9]
The power is in the downstroke.
Kautz et al. (1991)Turning force through the cycle
The same curve around the crank
14 elite cyclists · Lower workloadof positive crank work came from the downstroke. [17]
What the research shows
Almost all the positive work came from the downstroke.
At the lower workload, the downstroke supplied 98.6% of positive crank work in the group-average curve; the upstroke supplied 1.4%. At the higher workload, the downstroke share was 96.3%. Unweighting reduces opposing torque; positive upstroke work is a separate, small contribution. [17]
Pros reduce resistance on the way back up.
Among 11 professional, 13 elite and 14 club cyclists riding at 200–300 W, professionals had less resistive upstroke torque. A better return can help the other leg’s push; it does not establish zero upstroke power. [8]
The faster group pushed harder.
Nine elite-national riders produced 11% more power than six good-state riders in a one-hour test. Their higher power came primarily from larger downstroke forces. [9]
About the data
Both plots use the same illustrative curve to explain positive and negative crank torque. Their shape, amplitudes and areas are schematic, not a reconstruction of the study or a calculation of 98.6%. The moving markers show the same crank angle in each view. The separate 98.6% result comes from Kautz et al.’s lower-workload group curve: 14 elite male cyclists, approximately 61% VO₂max and 90 rpm.
Positive crank work measures mechanical contribution, not metabolic efficiency. The remaining 1.4% is positive upstroke work, not a measured efficiency gain or simply unweighting. These group results do not establish Bythlon-specific performance.
Read the paper ↗You don’t need to make a deliberate pull the goal of every pedal stroke. In Mornieux et al. (2008), coached pulling reduced net mechanical efficiency in both elite cyclists and non-cyclists at the tested submaximal workload. [2]
So, if actively pulling up is not a shortcut to efficiency in these tests, what else matters in a cycling setup?
Why cyclists still clip in
You don’t have to clip in to pedal efficiently. In steady-cycling tests, flat pedals with ordinary athletic shoes showed no significant efficiency disadvantage compared with cycling shoes and clipless pedals. [2] [12]
But energy cost isn’t the whole experience. A supportive platform, a comfortable foot position, and a connected feel matter too. You shouldn’t have to be locked to your bike to experience these benefits.
Clipping in showed no energy-cost advantage.
Straw & Kram (2016) · 10 cyclists analysed · 90 rpm
Metabolic power used (W)
50 W output
- Running shoes + flat pedals
- 445.7
- Running shoes + toe clips
- 428.7
- Cycling shoes + clipless pedals
- 441.6
100 W output
- Running shoes + flat pedals
- 619.0
- Running shoes + toe clips
- 600.7
- Cycling shoes + clipless pedals
- 612.3
150 W output
- Running shoes + flat pedals
- 817.9
- Running shoes + toe clips
- 818.0
- Cycling shoes + clipless pedals
- 806.4
No significant difference between setups was detected across these workloads (p = 0.57). [12]
Ten cyclists, seated cycling at 50–150 W and 90 rpm. Bars show mean metabolic energy use, rounded to 0.1 W; axes share the same scale.
That’s where Bythlon comes in. Our system pairs a rigid cycling shoe with a shaped cleat and matching pedal, designed for a secure, supported feel and comfortable movement without a mechanical lock. The cleat locates your foot on the pedal, float allows rotational movement, and you can lift away when you need to put a foot down.
More on shoe stiffness and comfort
Research on the foot–shoe–pedal interface reports changes in pressure, muscle activity and joint mechanics. Fletcher et al. (2019) found no overall gross-efficiency difference between two cycling soles of different torsional stiffness. Comfort and support are design considerations; these findings do not establish that stiffer soles always save energy or that a pedal system prevents knee injury. [10] [11] [13]
Norms evolve
Helmets offer a useful parallel in thinking about cycling safety: the evidence should match the claim. A systematic review and meta-analysis found that helmet use was associated with lower odds of head injury among cyclists involved in a crash or fall. [14]
For crit racers and velodrome sprinters putting in all-out efforts in short bursts, peak power is a different question from steady-state efficiency. A six-person sprint experiment found higher maximal power with toe clips than without them. The steady cycling results above should not be stretched into a claim about every type of riding. [15]
What about modern shoes and outdoor sprints?
Burns & Kram (2020) tested 12 male cyclists in 100-m uphill sprints. Toe clips increased maximal power relative to flat pedals with the same running shoes; cycling shoes with clip-in pedals increased it further. That is evidence about maximal sprint output, not steady-state energy cost. [16]
Our approach
Our pedal system offers a rigid shoe–cleat connection without the mechanical lock. The convex cleat engages the matching concave pedal body to locate the foot during the power phase. Built-in float allows rotational movement, while the open connection lets you lift your foot straight away.
Locate. Push. Lift away.
Bythlon mechanism · simplified illustration
- Locate
- Push
- Lift away
Product illustration based on Bythlon’s connection design. The shoe is omitted for clarity; arrows show motion, not measured force.
View the product demonstrationAt Bythlon, we believe you shouldn’t have to choose between performance and peace of mind. We hope to reduce cycling injuries by bringing safer performance cycling gear to the masses. We harness engineering, design, and biomechanics to empower every cyclist with confidence and joy, not fear.
Ride smarter, ride safer
Sources & study notes
-
Cycling research · 2007
Effect of pedaling technique on mechanical effectiveness and efficiency in cyclists
Korff, Romer, Mayhew & Martin
Medicine & Science in Sports & Exercise, 39(6), 991–995What it found: Deliberate pulling increased force effectiveness but reduced gross efficiency compared with the other techniques.
-
Cycling research · 2008
Effects of pedal type and pull-up action during cycling
Mornieux, Stapelfeldt, Gollhofer & Belli
International Journal of Sports Medicine, 29(10), 817–822What it found: Pedal type did not significantly change net mechanical efficiency; deliberate pulling reduced it.
-
Cycling research · 2009
Whole-body efficiency is negatively correlated with minimum torque per duty cycle in trained cyclists
Edwards, Jobson, George, Day & Nevill
Journal of Sports Sciences, 27(4), 319–325What it found: A more even torque pattern was associated with lower efficiency at some workloads.
-
Cycling research · 2010
Muscle coordination while pulling up during cycling
Mornieux, Gollhofer & Stapelfeldt
International Journal of Sports Medicine, 31(12), 843–846What it found: Pulling improved upstroke effectiveness while changing muscle activation and co-activation.
-
Cycling research · 2012
Effects of different pedalling techniques on muscle fatigue and mechanical efficiency during prolonged cycling
Theurel, Crepin, Foissac & Temprado
Scandinavian Journal of Medicine & Science in Sports, 22(6), 714–721What it found: Gross efficiency was lower with feedback during the first 15 minutes only. The feedback condition showed a smaller post-exercise maximal-power decline.
-
Research review · 2019
Strategies for improving the pedaling technique
Duc, Bertucci & Grappe
Journal of Sports Medicine and Physical Fitness, 59(12), 2030–2039What it found: Improved pulling mechanics had not consistently translated into significant gross-efficiency or performance gains.
-
Single-rider case study · 2021
Metabolic power and efficiency for an amputee cyclist: implications for cycling technique
Elmer & Martin
Journal of Applied Physiology, 130(2), 479–484What it found: Reducing the need to pull up lowered metabolic power and increased gross efficiency.
-
Cycling research · 2016
Differences in pedalling technique between road cyclists of different competitive levels
García-López, Díez-Leal, Ogueta-Alday, Larrazabal & Rodríguez-Marroyo
Journal of Sports Sciences, 34(17), 1619–1626What it found: Professionals showed less resistive upstroke torque and a higher positive impulse proportion.
-
Cycling research · 1991
Physiological and biomechanical factors associated with elite endurance cycling performance
Coyle, Feltner, Kautz, Hamilton, Montain, Baylor, Abraham & Petrek
Medicine & Science in Sports & Exercise, 23(1), 93–107What it found: The stronger group’s higher power came primarily from greater downstroke forces.
-
Research review · 2016
Intervention at the foot-shoe-pedal interface in competitive cyclists
FitzGibbon, Vicenzino & Sisto
International Journal of Sports Physical Therapy, 11(4), 637–650What it found: Interface changes can alter lower-limb mechanics; evidence for preventing or treating knee injury was limited.
-
Cycling research · 2019
The effect of torsional shoe sole stiffness on knee moment and gross efficiency in cycling
Fletcher, Asmussen, Nigg, MacIntosh & Nigg
Journal of Sports Sciences, 37(13), 1457–1463What it found: No significant overall difference in gross efficiency or three-dimensional knee moments between the shoe conditions.
-
Cycling research · 2016
Effects of shoe type and shoe–pedal interface on the metabolic cost of bicycling
Straw & Kram
Footwear Science, 8(1), 19–22What it found: No statistically significant shoe–pedal condition effect on metabolic power (p = 0.57).
-
Systematic review · 2018
Potential factors associated with knee pain in cyclists: a systematic review
Bini & Flores Bini
Open Access Journal of Sports Medicine, 9, 99–106What it found: Evidence connecting biomechanical factors with knee pain remained limited; the reviewed studies lacked prospective designs.
-
Systematic review & meta-analysis · 2017
Bicycle injuries and helmet use: a systematic review and meta-analysis
Olivier & Creighton
International Journal of Epidemiology, 46(1), 278–292What it found: Helmet use was associated with lower odds of head injury.
-
Cycling research · 1997
Torque-velocity relationship during cycle ergometer sprints with and without toe clips
Capmal & Vandewalle
European Journal of Applied Physiology and Occupational Physiology, 76(4), 375–379What it found: Maximal power was higher with toe clips in this experiment.
-
Cycling research · 2020
The effect of cycling shoes and the shoe-pedal interface on maximal mechanical power output during outdoor sprints
Burns & Kram
Footwear Science, 12(3), 185–192What it found: Attachment and the cycling-shoe/clip-in combination increased maximal sprint power in these comparisons.
-
Cycling research and public dataset · 1991
The Pedaling Technique of Elite Endurance Cyclists: Changes With Increasing Workload at Constant Cadence
Kautz, Feltner, Coyle & Baylor
International Journal of Sport Biomechanics, 7(1), 29–53What it found: The downstroke accounted for 98.6% of positive crank work at the lower workload and 96.3% at the higher workload. The upstroke generally changed from resistive torque to small positive or near-zero torque as workload increased.