Memorandum by the Transport Research Laboratory
ROAD TRAFFIC SPEED
The toll of death and serious injury resulting
from too much speed, and the effects on the quality of life, are
major issues of our time. Despite a very considerable body of
evidence showing the effect of speed on increased accident frequency
and severity, and the effects on communities, the problem remains
endemic. Our intention in this memorandum is to summarise in outline
the evidence from national and international sources and to submit
our views on how the problems associated with speed should be
When a 30mph speed limit was first introduced
in British towns in 1935, deaths fell by 15 per cent (Finch et
al, 1994). Between the 1950s and 1980s and onwards, many countries
introduced new speed limits, or lowered existing ones, with the
almost universal result that if they were enforced, speeds and
accidents both reduced (eg Salusjarvi, 1988; DSIR, 1963; Engel
& Thompson, 1988). TRL published results of a meta-analysis
of these studies (Finch et al, 1994), concluding from the
wealth of underlying data that, on average, each 1 mph reduction
in mean traffic speed is associated with a 5 per cent reduction
in accidents. In the US, increases in the national speed limit
have been shown clearly to be associated with increases in fatalities
(eg Garber & Graham, 1990).
A number of important studies (see Taylor et
al 2000 and Kimber 2001) have since addressed the relationship
between speed and accidents directly, as follows:
TRL studies of individual drivers
have addressed the question of how drivers' speed choices affect
their likelihood of being involved in an accident (Taylor et
al 2000; Quimby et al, 1999a; Maycock et al,
1998). The accident records of more than 10,000 drivers were related
statistically to their observed speeding behaviour, with the clear
result that accident risk rises the faster a driver travels, in
relation to those around him. At a speed of 25 per cent above
the average speed, the risk of accident involvement rises more
than 500 per cent. So the greatest reduction in casualties will
come from reducing the speeds of the fastest drivers.
TRL road-based studies addressed
the question of how the speed on a given road affects accidents
occurring there (Taylor et al, 2000). Several hundred thousand
observations of vehicle speed on more than a hundred roads of
different types were related statistically to the numbers of accidents
occurring on those roads. Again, the major impact that speed has
on accidents is clear: (i) the faster traffic moves on average
on a given type of road, the more accidents there are. Injury
accidents rise rapidly with average speed: approximately in proportion
to its squareif the spread of speeds (the ratio of the
standard deviation to average speed) remain constant. (ii) the
bigger the spread of speeds around the average speed, the more
accidents there are. The frequency of accidents rises rapidlyindeed
exponentiallyas the spread of speeds increases, for a given
average speed. (iii) these effects of speed on accidents depend
on the road type and situation.
Many "before and after"
studies of measures to slow traffic and improve safety have been
reported. These measures include: traffic calming schemes in 20mph
zones (Webster & Mackie, 1996)where injury accidents
more than halved; traffic calming in villages (Wheeler & Taylor,
2000)where injury accidents reduced by more than 20 per
cent; speed cameras (DTLR, 2001a)where injury accidents
reduced by a third; village speed limits (Suffolk County Council,
1999)where injury accidents reduced by 20 per cent. This
evidence for a link between reduced speeds and reduced accicdents
is very strong in itself.
Australian work examined speeds of
accident-involved cars and found them to be higher than speeds
of other cars, under the same conditions (Kloeden et al,
Together these studies provide robust and compelling
evidence of the nature and magnitude of the effect of speed on
accidents. They relate to real traffic on real (largely UK) roads.
They are large-scale studies involving rigorous statistical analysis.
Very importantly, they take account (Taylor et al, 2000)
of the complex interaction between the relevant variables, an
issue which, if ignored, can lead to misleading conclusions.
In the "before and after" studies
above, the more severe the accidents the greater the reduction
in their frequency. Swedish work (Andersson and Nilsson, 1997)
also shows that whilst injury accidents are proportional to the
square of average speed, accidents involving fatal/serious injury
are proportional to the cube of the average speed and fatal accidents
are proportional to the fourth power of the average speed. Other
studies show increased probability of injury with impact severity
(Hobbs & Mills, 1984). Higher impact speed means greater pedestrian
injury severity (DETR, 2000).
Illegal speed is speed in excess of the limit;
inappropriate speed is speed that is too fast for the prevailing
conditions (and may include illegal speed). Interpretations of
these phrases vary and care needs to be taken in what is meant.
Research relating to the distinct (ie separable) effects of illegal
and inappropriate speed in accidents is not so extensive, but
The identification of a relationship
between accident frequency and the proportion of vehicles exceeding
the speed limit (Taylor et al, 2000). This shows that:
firstly, the higher the proportion of drivers speeding, the more
accidents. If this proportion of speeders doubles, accidents go
up by 10 per cent. Secondly, the higher the average speed of those
drivers who speed, the more accidents: the frequency of
accidents rises by 19 per cent if the average speed of the speeders
increases by one mph. Again the potential to reduce casualties
by reducing the speed of the fastest drivers is clear.
data based on police records collected at accident scenes also
provide some evidence about the role of inappropriate speed. These
enable the proportion of accidents to be estimated in which inappropriate
speed was deemed to be a factor (Broughton et al, 1998).
The concept relies on the subjective judgement of a police officer
attending the accident scene after the event, of the "appropriate
speed" in a given situation, and of its connection with other,
related factors. Whilst it is clear from these studies that inappropriate
speed is a factor in a substantial proportion of accidents, the
exact magnitude of this proportion is difficult to quantify. In
contrast, the objective evidence relating directly to the speed-accident
relationship (above) is demonstrably more relevant to: (i) quantifying
the influence of speed in different situations, (ii) gaining a
real understanding of the issues, and (iii) addressing the range
of very different problems that exist.
Most of the now extensive evidence relates to
illegal speed. The AA Foundation (2000) suggests that this is
what most people think of as being inappropriate. Quimby et
al (1999a) found in a study of more than 5,000 drivers that
site characteristics have by far the biggest influence on drivers'
speed choice. Speed choice is further influenced by personal characteristics,
which include demographic and psychological characteristics (Webster
and Wells, 2000). In the former category, mileage travelled has
been identified as a key factorhigher mileage drivers tending
to be faster drivers (eg Quimby et al, 1999). The influence
of psychological characteristics is well-documented (eg, AA Foundation,
2000; Corbett, 1999; Stradling, 1999; Quimby et al, 1999a,b).
Non-UK sources are considered less relevant because of the differences
in culture, legislation and enforcement between countries.
Attitudes to speeding need to be changed (eg
see AA Foundation, 2000) and road environments need to better
explain the appropriate speed in different situations (eg DETR,
2000). A greater understanding is required of how to encourage
people to understand/accept the need to reduce speed.
There is a body of research considering factors
that influence pedestrian casualty levels, and solutions (eg,
Taylor et al, 1996; Summersgill et al, 1996 and
others who have identified the determinants of pedestrian accidents
on road links and at junctions of different types, and AA Foundation,
1994, which relates risk to road/housing type). Speed is not directly
considered in these studies, but Taylor et al (2000) show
that both pedestrian and vehicle accidents are reduced with lower
speeds. Urban Safety Management principles centre on diverting
traffic away from streets with high risk to pedestrians and slowing
the remaining traffic (IHT 1990).
A study of child pedestrian accidents in Europe
(MVA Limited, 1999) suggests that the lower incidence of low speed
limits in the UK, and of measures to reduce speed, contribute
to the higher pedestrian casualty rate here. Such rates are particularly
marked amongst low income and ethnic groups. Other key reasons
for this elevated rate result from differences in the type of
road networks, and adjacent housing, in the UK compared to other
The quality of life is usually taken to include
both objective measuresinjuries as discussed above, noise,
health and fitnessand the more subjective social issues.
The effects of speed are not distributed evenly across society
as a whole. The impact on risk for example, per unit of exposure,
is greater in some groups than in the population at large. This
bears particularly on children, whose cognitive development, including
the judgement of speed, is less advanced (Ampofo-Boateng and Thompson,
1991; Sandel 1973) and on elderly people whose declining physical
capacity and lower concentration skills make them particularly
vulnerable (eg Hine and Russell, 1996; AA Foundation, 1995).
The speed of vehicular traffic is an important
deterrent to cycling (eg Bracher 1989) and, it is widely believed,
walking. Since physical activity is important in reducing heart
disease and strokes this has health consequences. These have to
be balanced against the increased risks of cycling and walking
compared with driving. Better knowledge is needed about the possible
effect of traffic speed in suppressing walking activity.
Long streams of fast traffic can cause social
exclusion in communities by severance, and by making it difficult
for those without cars to achieve adequate mobility (DETR, 2000).
There is evidence of perceived improvements in the quality of
life where lower speeds have been engineered through traffic calming
(eg Danish Roads Directorate, 1987; DTLR, 2001b; Webster, 1998).
Home Zones, in which road space is shared between
motor vehicles and other road users, should impact on local residential
urban design in the future. These change the functional use of
the streets and engineer more appropriate vehicle speeds. However,
the proportion of urban space appropriate for this form of treatment
Inappropriate and excessive speed are widespread
and a range of measures is needed to address them. At present
conflicting messages and cues are given to drivers from different
elements of the transport system (through road design, speed limit
allocations, and vehicle capabilities) and from different practices
adopted by Highway Authorities. It will be extremely important
to move effectively to a more coherent set of measures, and to
reduce these conflicts.
Existing limits are in practice applied locally
using different criteria (AA Foundation, 2000). The criteria for
both built up and rural environments need to be rationalised to
consistent national standards. This requires simpler procedures
for changing speed limits as conditions change over time, whilst
ensuring that, at each change, compatible speed limits, road design
and traffic behaviour are achieved. There is substantial evidence
of the lack of effect of limits introduced into inappropriate
road environments (Finc et al, 1994; Mackie, 1998). But
there are also examples where lower speed limits alone have been
introduced with apparent benefits in reduced accidents: in Suffolk
villages and in Scottish trials; but in both cases extensive publicity
or commitment by the local community was required.
The speed limit regime adopted in some other
European countries (notably the Netherlands) is generally lower
than in Britain. The norm for roads other than high quality roads
is 80kph (50mph). We believe that a similar system relating speed
limits closely to road quality should be adopted in Britain. This
would require roads to be allocated to speed limit groups based
on the accident risk for that type of road in relation to the
speed and other mobility and environmental criteria. If an existing
road is not up to the required safety standard for the design
speed then a lower speed should be adopted until the road can
be improved to the appropriate standard. Roads chosen to fulfil
a lower traffic function should be designed for lower speeds accordingly.
To initiate this process it would be necessary
to assess the current speeds and design of all GB roads. A programme
is already underwaythe European Road Assessment Programmesupported
by the motoring organisations and several national governments
including the UK government, which aims to develop road risk assessments
on a common basis nationally and progressively across Europe,
and clearly associate risk with road type in drivers' perceptions.
Urban Road Redesign
20mph zones have achieved considerable success
in residential areas in reducing accidents through reduced speeds
(Webster and Mackie, 1996). But the more major roads in towns
usually need to cope with a wider variety of conflicting functions,
requiring treatments to allocate space between competing modes
and road users. Completely segregated activities are usually not
feasible. Therefore the road design and speed limiting measures
need to take account of the risk of the most vulnerable groups.
(See Mackie, 1998 and DETR, 2000). The Gloucester safer City project
(DTLR, 2001c) provides a good example of how a range of speed
reducing measures is essential in wider urban safety management
Rural Road Measures
Traffic calming measures within villages, including
"Gateway" treatments (Wheeler et al, 1994, 1999)
have been used to reduce speeds with the consequence of reduced
accidents (Wheeler and Taylor, 2000). Innovative perceptual measures
(including roundels and "countdown" markings) have also
been trialled with varying success (Barker and Helliar-Symons,
1996; Barker, 1997). Interactive signs can be useful in encouraging
lower speeds at bends and junctions on rural roads, and are effective
in targeting the higher speed drivers (Winnett et al, 1999).
Measures for Pedestrians and Cyclists
The accident risk per unit of exposure for pedestrians
and cyclists is substantially higher than for car occupants. Whilst
complete physical segregation of these modes is usually not possible,
it can radically reduce the risk. In most circumstances where
full physical segregation is not feasible, it is extremely important
to develop solutions for cyclists and pedestrians which reduce
their elevated risks. This we see as an essential pre-requiste
to achieving safe increases in cycling and walking, and the health
benefits that should result.
The Potential of New Technology; Variable Speed
Technology is available currently that, if sufficiently
reliable, could force drivers to adopt lower speeds. The potential
acceptability of such a measure raises political issues. Compulsion
is most likely to be effective if it is targeted on a minority
who fail to comply with generally accepted behaviour, rather than
as a tool to change the behaviour of the majority. Shifts in majority
behaviour are best effected by engineering and attitude change.
However, technology can provide increasingly effective means of
presenting information and advice to drivers whilst on the move
(for example from in-vehicle devices signalling the speed limit).
Such a stream of development, whilst longer-term, could produce
large accident savings.
Redesign of Vehicle Fronts
Injuries to pedestrians struck by cars can be
considerably reduced by redesigned car fronts. The optimum testing
process to discriminate between those cars giving better protection
and those giving worse protection than average has now been rejected
by EC, in favour of a voluntary code of practice by car manufacturers.
From an examination of the types of injury that might be mitigated
under the voluntary system, in relation to the impact speed, we
believe this will provide only a quarter to a third of the benefits
in fatal accident savings of an EC directive based on the best
available test methods. The potential benefits achievable by better
car front design are most at impact speeds of around 20 mph. This
re-emphasises the importance of reducing the higher speeds in
Many drivers have difficulty in recognising
the added risk from increased speed. Mis-perceptions abound. In
many respects attitudes and beliefs are similar to those on drinking
and driving some 40 years ago. The change in behaviour on drinking
and driving which resulted in very large reductions in death and
injury came from a series of measures following from evidence
produced in the 1960s which clearly showed the relationship between
blood alcohol level and accident risk. Now, in 2002, we have a
very considerable body of evidence showing the relationship between
speed and accident risk.
This should now be used as the trigger to a
process of changing attitudes, driven by a sequence of measures
and messages, each one picking up and carrying forward the attitude
shift achieved by the previous one. The measures need to be coherent,
and to generate a clear perceptual understanding by the driver
of the risk inherent in the speed they adopt for a particular
The foregoing underlines the importance of coherence
and consistency in the measures adopted by all of the relevant
bodies. The essential elements needed for coherent policy and
implementation on speed management exist within the system of
Local Transport Plans, and the statutory responsibilities of Highway
Authorities, including particularly the Highways Agency for the
national network, and the Local Highway Authorities for local
networks. But speeding, and speed management, are complex issues,
and the establishment of clear frameworks for action is still
lacking. We should not underestimate the importance of clear and
convergent practices in speed management; and we are not convinced
there is yet a sufficient single driver of change.
The substantial body of knowledge now available
on the effects of speed and the factors influencing it, place
government in a strong position to drive the change required.
We believe this is now a pressing need.
There are in the meantime two particular issues
that need to be addressed.
The use of Speed Cameras
The police should give most emphasis to targeting
those speeding offenders whose speed has most effect on accident
risk. The objectives of enforcement should be clarified. If there
is a very localised speeding problem, then the use of a well-marked
camera can be effective in highlighting the local risk. But a
key role of enforcement should also be as part of a broader strategy
to modify drivers' speed choice to levels which properly reflect
accident risk everywhere. In this context, cameras should be used
more generally so as to remind drivers of the consequences of
adopting excessive speeds. There is no evidence that painting
cameras bright yellow will have any effect in additional accident
reduction. Moreover, highlighting these cameras through the use
of conspicuous colours gives a counter message, that may tempt
drivers to adopt higher speeds generally in the confidence that
they can brake hard when they spot a camera.
Sentencing and Penalties
A forthcoming DTLR (2002) report shows excessive
speed to be the most common element among dangerous driving cases;
but notes a lack of consistency between agencies in how speeding
is used to determine dangerous driving. Police clearly feel CPS
guidelines are insufficient in relation to speed (and other offences):
magistrates feel guidelines are better for speeding than for some
other offences. The Home Office review of penalties proposed a
two level fixed penalty for those exceeding the speed limitwith
a higher penalty for drivers exceeding the limit by more. This
is consistent with the effect of speed on accidents shown by TRL
research. It also proposed updating penalties over time, which
would further support campaigns to reduce speeds.
This summary presented in this Memorandum, is
of necessity, very brief and rests mainly on the evidence cited
in the references.
Taken together, the evidence showing the damaging
effects of excessive and inappropriate speed on both the frequency
and severity of accidents and on the quality of life is extensive
and compelling. We believe that there is a pressing need for actions
to reduce such speeds.
Transport Research Laboratory
7 January 2002
Ampofo-Boateng K and Thomson J (1991). Children's
perception of safety and danger on the road. British Journal of
Psychology, 82, pp 487-505.
Andersson G and Nilsson G (1997). Speed management
in Sweden. Swedish National Road and Transport Institute VTI,
Barker J (1997). Trials of Rural Safety Measures.
TRL Report 202. TRL Limited, Crowthorne.
Barker J and Helliar-Symons R D (1996). Count-down
signs and Roundel marking signs. TRL Report 201. TRL Limited,
Bracher T (1989). Policy and provision for cyclists
in Europe. Commission of the European Communities, Directorate
General for Transport, Brussels.
Broughton J, Markey K A and Rowe Supt D (1998).
A new system for recording contributory factors in road accidents.
TRL Report 323. TRL Limited, Crowthorne.
Corbett C (1999). What would stop drivers speeding?
Proc. PACTS conference Speed: Whose business is it? 10 February,
Danish Roads Directorate (1987). Consequence
evaluation of environmentally adapted through road in Vinderup-Report
52, Herlev: Road Data Laboratory.
DETR (2000). New directions in speed managementa
review of policy. London.
DSIR (1963). Research on Road Safety. Department
of Scientific and Industrial Research: Road Research Laboratory.
DTLR (2001a). Cost recovery system for traffic
safety camerasFirst Year ReportExecutive Summary.
DTLR Road Safety Division, London.
DTLR (2001b). Urban street activity in 20mph
Zones. DTLR, London.
DTLR (2001c). Gloucester Safer City. DTLR, London.
DTLR (2002 in preparation). Dangerous Driving
and the Law. DTLR RSRS 26 DTLR, London.
Engel U and Thomsen L K (1988). Hastigheder,
hastighedsgraenser og ulykker. Report 27. Centre for Traffic Safety
Finch D J, Kompfner P, Lockwood C R and Maycock
G (1994). Speed, speed limits and accidents. Project Report PR
58. TRL Limited, Crowthorne.
Garber S and Graham J D (1990). The effects
of the new 65 mph speed limit on rural highway fatalities: a state-by-state
analysis. Accident Analysis and Prevention.
Hine J and Russell J (1996). The impact of traffic
on pedestrian behaviour: assessing the traffic barrier on radial
routes. Traffic Engineering and Control, February, pp 81-85.
Hobbs C A and Mills P J (1984). Injury probability
for car occupants in frontal and side impacts, TRRL Report 1125.
TRL Limited, Crowthorne.
Institution of Highways and Transportation (1990).
Guidelines for Urban Safety Management.
Kimber R M, (2001). What role does speed play
in crashes? Invited address to the Belgian Presidency of the
Council of the European Union : Killing Speeds-Saving Lives.
Kloeden C N, McLean A J, Moore V M and Ponte
G (1997). Travelling speed and the risk of crash involvement.
Federal Office of Road Safety, CR172. Canberra.
Mackie A (1998). Urban speed management methods.
TRL Report 363, TRL Limited, Crowthorne.
Maycock G, Brocklebank P J and Hall R D (1998).
Road layout design standards and driver behaviour. TRL Report
332. TRL Limited, Crowthorne.
MVA Limited (1999). Comparative study of European
child pedestrian exposure and accidents. Woking.
Quimby A, Maycock G, Palmer C and Buttress S
(1999a). The factors that influence a driver's choice of speeda
questionnaire study. TRL Report 325. TRL Limited, Crowthorne.
Quimby A, Maycock G, Palmer C and Grayson G
(1999b). Drivers' speed choice: an in-depth study. TRL Report
326. TRL Limited, Crowthorne.
Salusjarvi M (1988). The speed limit experiments
on public roads in Finland. Proc. Road and Traffic Safety on two
Continents. VTI Rapport 332A Vag-och Trafik Instituet, Linkoping,
Sandel S (1973). Children and traffic. London:
Stradling S (1999). Why drivers speed. Proc.
PACTS conference Speed: Whose business is it? 10 February, London.
Suffolk County Council (1999). Suffolk 30mph
speed limits. Press Release, 12 October 1999.
Summersgill I and Layfield R E (1996). Non-junction
accidents on urban single carriageway roads. TRL Report 183. TRL
Taylor M C, Hall R D and Chatterjee K (1996).
Accidents at 3-arm traffic signals on urban single-carriageway
roads. TRL Report 135. TRL Limited, Crowthorne.
Taylor M C, Lynam D A and Baruya A (2000). The
effect of drivers' speed on the frequency of road accidents. TRL
Report 421. TRL Limited, Crowthorne.
The AA Foundation for Road Safety Research (1994).
Pedestrian activity and accident risk. Basingstoke.
The AA Foundation for Road Safety Research (1995).
Risk and safety on the roads: the older pedestrian. Basingstoke.
The AA Foundation for Road Safety Research (2000).
What limits speed? Factors that affect how fast we drive. Basingstoke.
Webster D C (1998). Traffic calmingpublic
attitude studies: a literature review. TRL Report 311. TRL Limited,
Webster D C and Wells P A (2000). The characteristics
of speeders. TRL Report 440. TRL Limited, Crowthorne.
Webster D C and Mackie A M (1996). Review of
traffic calming schemes in 20 mph Zones. TRL Report 215. TRL Limited,
Wheeler A H and Taylor M C (1999). Traffic calming
on major roads: Final report. TRL Report 385. TRL Limited, Crowthorne.
Wheeler A H and Taylor M C (2000). Changes in
accident frequency following the introduction of traffic calming
in villages. TRL Report 452. TRL Limited, Crowthorne.
Wheeler A H, Taylor M C, and Barker J (1994).
Speed reductions in 24 villages: details from the VISP study.
Project Report PR85. TRL Limited, Crowthorne.
Winnett M A, Woodgate E, and Mayhew N (1999).
Interactive Fibre Optic signing at a rural crossroads. TRL Report
TRL 401. TRL Limited, Crowthorne.