This list is intended to be without bias, but little evidence has been found to suggest that cyclists are safer on paths than on roads. If you know of other research, please contact John Franklin.
References are listed by order of date, then title.
Transport Advisory Council, Ministry of Transport, UK, 1938.
Recommendations to improve the safety of cycling with regard to cycle tracks, lighting, careless riding, insurance, etc.
Cycle tracks increase danger at every road junction. Supports trend towards merging tracks into carriageway before cross-roads. Considers cycle tracks provide safety benefit between junctions (but provides no evidence and disputed by one Council member). Tracks should be of adequate width, reasonably continuous and properly surfaced. Cycles with three or more wheels should be prohibited from using tracks.
Supports principle of making it compulsory for cyclists to use cycle tracks (but records minority dissent to this view in the Council), but this would be inappropriate where cycle use is high (cycle tracks can't cope with high volumes of cyclists) and sometimes at night.
Kaplan. FHWA, US, 1975.
Full paper available
on line
National survey of LAW members.
Cycle paths 292 accidents per million
cycle miles, against 104 for minor roads and 111 for major roads.
Beukers. Dutch Ministry of Transport. ITE Conference, 1977
Safety problems of two-way cycle tracks at junctions almost insuperable.
Rainbird. Hertfordshire County Council/PTRC, UK, 1979.
98 cycleway accidents in official (Stats 19) statistics during 5-year period. 82 involved mopeds (which can use the paths), 40 involved cyclists and 20 involved pedestrians. 43% were at cycleway junctions.
Principal accident types were collisions at road junctions and underpasses (poor visibility a common factor), head-on collisions between users and loss of control. Also incompatibility of pedestrians and cyclists/mopeds on paths. Lack of user discipline a major problem and likely to be difficult to correct.
First, and most important, is the need for consistency in layout, signing and markings. Unlikely to be a second chance to get cycleway design right.
Addendum: In early 1980s Stevenage cycleways averaged 20 reported accidents a year despite remedial measures.
Alrutz, HUK-Verband, Köln, Germany, 1980.
Presented at Velo City,
Bremen, 1980 and in conference proceedings.
Based on study of 4,000 accidents in Köln 1976 - 1978. Cycle paths as traditionally built do not guarantee a reduction in casualties. The risk cyclists face depends on how often their unimpeded ride is interrupted.
Where paths exist, 41% of crashes on links happen at driveways, 10% are related to stationary vehicles, 17% as cyclists leave the path. Where no paths exist, crashes due to interruptions are mostly due to parked vehicles.
On roads with cycle paths on both sides, such cyclists are involved in one-third of crashes, although they account for only 20% of traffic. Cyclists using a path in the counter-flow direction are at above average risk. Between junctions, accidents on the carriageway may be more serious than on cycle paths.
More than one-half of accidents to cyclists occur at junctions. 80% are related to turning against or crossing the path of other vehicles. Where cycle paths are present, the most common collision type involves motor vehicle turning right and cyclist going ahead. Often the motor vehicle fails to cede right of way, but at traffic lights cyclists in most cases passed at red.
Linderholm. University of Lund, Sweden, 1984
Reference to study at Salzburg conference
7,500 cyclists studied.
At junctions cycle tracks 3.4 times more dangerous than using road, but
rising to 11.9 riding in 'wrong' direction.
Cycle lane 1.1 times more dangerous than unsegregated road for going ahead.
Left turns particularly dangerous from cycle track (11.0), less so from
cycle lane (3.4).
Proportion of mistakes highly noticeable in cyclists making left-turn from
one-way cycle track. Cycle track users most likely to run red lights or cycle on
pavements.
Poor co-operation between road users where there is a cycle track.
Der Polizeipräsident in Berlin. Berlin Police, Germany, 1987.
Berlin Police study 1981 to 1985.
Cyclists 4 times more likely to have accident on roads with cycle paths.
Likelihood of serious or fatal injury similarly increased.
Full paper and English translation
Bach, Rosbach, Joergensen. Vejdirekforatet, Denmark, 1988
Tabulated results
Traffic safety of cycle tracks in Danish cities.
Before and after study of 105 new cycle paths in Denmark, introduced
1978-81, totalling 64km. Cyclist casualties increased 48% following
introduction of paths. Car drivers, moped riders and pedestrians also suffered
more accidents, with overall rise in casualties of 27%.
Mills. TRL RR220 UK, 1989.
Study based on Oxford, John Radcliffe Hospital.
20 accidents (3%) occurred on cycle tracks, 90% not reported.
50 accidents (7%) on footways, 98% not reported. 18 accidents (2%) on
footpaths, none reported.
594 accidents (81%) on roads, 64% not reported.
52% of accidents involved no other vehicle.
Comment: Cycle track accidents high relative to mileage and use of tracks in catchment area.
Leden. Technical Research Centre of Finland, 55 (ISBN 951-38-3436-0). 1989.
Field survey of 14,000 schoolchildren between ages of 6 and 16 in five
Swedish, one Norwegian and three Finnish towns. Children described road crashes
in which they had been involved over the past year.
For children cycling,
risk of colliding with motor vehicle 2.7 times higher at intersections with a
cycle track (which the child used) than at road-only intersections. Risk highest
when cycle crossings 8 to 15m from intersections and when traffic signals were
present.
Overall risk of collision is 0.5 crashes/100,000km on the
carriageway but 1.3 crashes/100,000km on a cycle track, rising to 2.8 when there
are concurrent green signals for road and cycle track at junctions.
Review in Roue Libre, 1991.
Review of new Belgian Government policy on cycling.
Cycle tracks no longer favoured in urban areas due to problems and danger.
Preferable to remove narrow paths and where in bad state.
Two-way cycle tracks to be declassified as particularly dangerous.
Schmidt. Proceedings of ICTCT Workshop. Bulletin 110, University of Lund.
1991.
Referenced from Safety of vulnerable road users.
Study based on Vienna. Conflicts generated mainly by narrow footpaths and cycle tracks, (relatively) high speed of cyclists, poor visibility, considerable age difference.
Hass-Klau et alia. Environmental & Transport Planning, UK/Germany 1991.
Anglo-German comparison; many German references.
Number of motor vehicles and in particular number of cyclists has much
stronger influence on safety than cycle facilities. Some main roads with cycle
facilities have higher cycle accident rate than without. 36% of accidents in
German study towns take place when the cyclist is using a facility. More serious
accidents a result of cross-manoeuvres or with no other vehicle at all.
Visibility and care crucial; cycle facility may contribute to accidents by
making cyclist over-confident. Facilities cause many problems; bad cycle
facilities are worse than none.
Peterborough: high accident rate in residential areas casts doubt on
independent cycle facilities. York and Oxford: high serious accident rates.
Preferred policy of cyclists is better junction design and safety
education.
Ketteridge. Milton Keynes Development Corporation, UK, 1991.
Includes one-month hospital survey which showed 14 cycle path accidents against 1 minor road accident and no major road accidents in equivalent area. All 3 serious accidents were cycle path. A further 15 cycle path accidents were not explained (possibly cycle/pedestrian).
Wegman, Dijkstra. SWOV, Netherlands, 1992.
Originally presented to Roads
and Traffic 2000 conference, Berlin, 1988;
Revised version included in Still
more bikes behind the dikes, CROW, 1992.
In built-up areas cycle tracks 25% safer than unsegregated road between
junctions, but 32% more dangerous at junctions. Cycle lanes 36% more dangerous
between junctions, 19% safer at junctions. Seriousness of accidents greater if
tracks or lanes present compared with no facilities. Cycle lanes narrower than
1.8m particularly hazardous.
Outside towns, cycle track safety depends on car and cycle numbers.
New cross-town routes in Den Haag and Tilburg had produced no safety gain
and had not encouraged much new cycling.
ARGUS, Austria, 1992.
Report of Velo Secur cycle safety conference in Salzburg.
Translation of conference summary
Trying to keep cyclists apart from motor traffic in urban areas has proved
too much for road users to cope with and led to unnecessary accidents. Urban
cycle paths are unsuitable and should not be used.
Providing cycle paths
rarely solves safety problems and often introduces new ones. Austrian and
Swedish research shows cyclists at 3 times greater risk on cycleways than on
road.
Bremen police chief agreed and called for compulsory use of paths to
be lifted.
Harland, Gercans. TRL, UK, 1993.
Analysis of experimental routes in Exeter, Bedford, Nottingham and Stockton.
No evidence that cycle routes lead to more cycling or improved safety. No
change in Stats 19 (reported) casualties, although some cyclists transferred to
paths.
Comment: It is probable that there was an overall increase in accidents due to the transfer of some cyclists to paths where the rate of reporting is much lower. Other evidence from Bedford gives weight to this.
Schnull, Alrutz et al. German Federal Highways Institute Report 262, 1993.
Proportion of junction accidents significantly higher with cycle tracks. HGV conflicts more common with segregation. Without signals, cyclists nearly 5 times more at risk on a cycle track; contrasting surfaces only reduces this to 1.5. With signals cyclists 1.7 to 2.7 times more at risk on cycle track, 1.3 times on a cycle lane. At roundabouts cycle tracks increase risk by 30%, cycle lanes by 25%.
Ahlcrona et al. Sweden, 1994.
Referenced from Safety
of vulnerable road users.
Small sample investigation of crashes to vulnerable road users at Skåne. Two-thirds of collisions occurred at pedestrian and cycle track crossings. In two-thirds of the cycle crashes, the cyclist was riding on a two-way cycle path counter to traffic flow or appeared somewhere he was not supposed to be.
Wachtel, Lewiston. ITE Journal, USA. September 1994.
Full paper
available on line
Sidewalks or paths adjacent to a roadway are usually not, as non-cyclists expect, safer than the road but much less safe. Risk on average 1.8 times as great, but much greater for wrong-way sidewalk bicyclists. Sidewalk cycling appears to increase incidence of wrong-way travel on both paths and roads. Separation of bicycles and motor vehicles leads to blind conflicts at intersections. Aim of well-designed roadway system should be to integrate bicycles and motor vehicles, according to the well-established and effective principles of traffic law and engineering, not to separate them.
Danish Road Directorate. Traffic Safety and Environment Report 10, 1994.
Referenced from Safety of vulnerable road users.
Bicycle paths and lanes offer overall improvement in safety on links. However, cycle lanes result in more crashes with parked cars. Cycle tracks cause major safety problems at signalised junctions. Better to replace track with cycle lane 20-30m before junction.
Larsen. Journal of Traffic Medicine, 1994.
Referenced from
Safety of vulnerable road users.
44 per cent of cyclist collisions in Denmark occur to cyclists using a cycle track. 42 per cent to cyclists using the normal carriageway.
Eilert-Peterson, Schelp. Pergamon, Sweden, 1995.
12-month study of Västmanland county based on emergency visits to
physician or dentist.
Cyclists were mostly injured on pavements, pedestrian malls and cycle
tracks.
20% of events occurred on urban roads.
Most common injury event involved no other party but due to environmental
or behavioural factors.
Räsänen, Traffic Safety Committee of Insurance Companies, Finland,
1995.
Referenced from Safety of vulnerable road users.
Study of 234 bicycle crashes in four Finnish cities. 63% of collisions between a cyclist and a motor vehicle took place at cycle track crossings. Such facilities are not well understood by motorists and may induce careless behaviour on the part of the cyclist. Children and elderly cyclists most at risk.
Morgan. TRL/Institute of Civil Engineers, UK, 1995.
Only 3% of injury accidents on cycle tracks and off-road are recorded.
Cycle facilities not improving use or safety. What we are doing now is
either insufficient or just plainly wrong.
Summala et al. Accident Analysis and Prevention, 1996.
Referenced from
Safety of vulnerable road users.
Analysis of motorist visual scanning when crossing a cycle track and preparing to turn at an adjacent junction. High crash risk when motorist turning right and cyclist comes from right. Compared with the normal situation at a junction, that road users have found a strategy to handle, the abnormal situation in the presence of a cycle track leads to a dangerous outcome.
Jensen, Andersen, Nielsen. Velo City, Barcelona, 1997.
Cycle lanes in Denmark realise accident savings between junctions, but this is more than outweighed by additional accidents at junctions. Cyclists particularly vulnerable at non-signalised junctions where study indicates a nearly fourfold increase in risk. At signalised junctions the marking of a coloured lane through the junction can assist safety.
Moritz. Transportation Research Board, US, 1998.
Paper available on-line.
Survey of 2,000 cyclists, update on 1974 survey.
Includes facility relative danger index. Major road 0.66, Minor road 0.94,
Recreational multi-use trail 1.39, footway 16.34.
Also details of 1994 Washington study, with similar results.
Leden, Gårdner, Pulkkinen. Transportation Research Board, 1998.
Swedish study based in Göteborg.
Conventional cycle tracks increase
cyclists' risk at junctions. Studied area had 126 reported accidents in 5 years
on 18.7km of track; many involved serious injuries. 47% no other vehicle, 10%
cycle/pedestrian, 7% cycle/cycle.
During study period 48 of 287 accidents were serious. 50% cycle/ped
accidents serious (for both users).
Raising crossings helps, but only 10% to 50% reduction in risk. Speed of
cyclists must be kept low for success. If cyclists go faster (i.e. they
ride at typical road speeds), crashes increase.
Franklin. Milton Keynes Cycle Users Group, UK, 1998.
Full report available on-line.
For later data, see: Two decades of the Redway cycle
paths of Milton Keynes.
History, use, safety and suitability of cycle
paths in Milton Keynes.
Includes review of last 10 years' crash statistics based on Police and
Hospital records and other sources. 6 deaths to cyclists off-road in 10 years
against just one in comparable area on roads.
OECD (Organisation for Economic Co-operation and Development), 1998.
Full report on line
A major analysis of safety for cyclists and pedestrians in OECD member countries, in the context of encouraging more people to walk and cycle. Looks at regulations, modal split, casualty statistics and causes. Top priorities are to change the social climate to favour pedestrians and cyclists by giving these modes more priority; measures to reduce speeds and promote traffic integration, and making roads 'legible' and the traffic environment 'forgiving'.
Limitations of cycle tracks and lanes recognised. In Denmark over 50% of cycle crashes occur on cycle tracks and lanes. Most severe and fatal crashes take place at road junctions and cycle track crossings. Accidents more severe when cyclist is hit by a turning vehicle. Cyclists find it difficult to scan adequately using a cycle crossing.
Other references from this report listed separately.
Aultman-Hall, Adams. Transportation Research Board, 1998.
Details of Toronto and Ottawa commuter surveys of comparative risk. In each
case roads safer than cycle paths, which are safer than footways. Footway
cyclists at greater risk wherever they ride.
Analysis of who uses footways.
Aultman-Hall, Kaltenecker. Transportation Research Board, 1998.
1995 study of roads, paths and footways used by commuters.
Cycle paths significantly more hazardous than roads and footways much more
so. Injury likelihood 10:20:70 roads:paths:footways.
Coates. Velo City, Graz, 1999.
Study of Oxford, UK, where 25km of cycle lanes have been introduced in last 20 years. Some accident savings along cycle lanes away from junction, mainly due to reduction in traffic speed as a result of the lanes taking space from motor traffic. At junctions, cycle lanes can increase accidents, especially if the lanes are not carried through the junction. Main benefits lie in reducing perceived risk, and assisting cyclists to pass other traffic in congested urban centres.
Franklin. Traffic Engineering + Control, 1999.
Full
paper available on-line.
Analysis of the UK's largest purpose-built cycle path network and the
consequences for cycle use and safety over two decades. Network has suppressed
rather than encouraged cycling and has proved to be consistently less safe than
the town's unrestricted main roads.
Injury accidents per million km cycled:
main roads 31, local roads 149, cycle paths 166. All crashes: main roads 47,
local roads 149, cycle paths 319.
Pasenen, Helsinki City Planning Department
Full paper
available on-line
In Helsinki, using a road-side cycle path is nearly 2.5 times likely to result in injury than cycling on the carriageway with traffic. At junctions the relative risk rises to more than 3 times. In those countries and cities which are just beginning to build cycling facilities, two-way cycle paths in particular should be avoided in an urban street network.
In Helsinki and Lund (Sweden), cycling leads to more pedestrian injuries per kilometre reported to the police than motor traffic.