It is easy to say that flying is safe; but safe compared with what? Fear of flying is only partly rational, which makes it difficult to persuade the afflicted with the unfeeling logic of statistics. Even when nervous fliers are provided with a comparison that brings the truth of flight safety into easy perspective, the ultimate hurdle is man's fear of falling from heights. The latter has never been reduced - let alone eliminated - by pointing out that people don't often fall to their death.

Those frightened of flying will hardly have been reassured by a grim procession of disasters in the 1990s, among them the TWA jumbo explosion off Long Island, New York. Worse still, the fundamental cause of this catastrophe may never be known. The likelihood of solving similar mysteries in future looks sure to be increased by the recent decision of the US safety authorities to demand that all airliners carry sophisticated flight data recorders, the 'black box'.

Despite recent events, flight safety should be put in statistical perspective. During 1999 the world's air travellers made just over 1.5 billion individual journeys on scheduled airlines. It was an average year for flight safety, compared with the last tenyears: there were 48 fatal accidents to civil airliners of all kinds, including the domestic short-hop propeller-powered type and cargo flights. The world total of airline deaths was low at 730; the average number of fatalities each year is just under 1,200.

Given the world average, a traveller would have to take 1.5 million flights before facing his or her statistical end. If that means very little to you, read on to put it into perspective. If it sounds horrifyingly dangerous, read on to discover how you can improve your chances enormously by knowing how to be selective about airline safety.

In the average fatal accident, more than half the people on board survive. It has also been shown that frequent air travellers have a better chance of surviving accidents than occasional travellers: this is assumed to be because they know the aeroplane better, panic less, and so can get out faster. Since you will want to be one of those who survives any accident that your flight has, take the emergency procedures briefing seriously. This is not paranoid, it's pure sense. Look at where all the exits are relative to you and imagine finding your way to them in the dark. Count the seat rows to them. Read the emergency cards carefully, study the brace position, have your seat belt firmly (really firmly) fastened at take-off and landing, and slacken it in flight but always have it fastened. Look with particular care at the diagram showing how to open the exit doors, and imagine opening them yourself in the dark. Having done all this, sit back and enjoy your flight.

Airlines specialise in delivering travellers over long distances fast and safely. Risk does not increase with distance on an airline flight, whereas it increases almost directly in proportion to the distance travelled in a car. According to statistics there is no country in the world where the average car driver could expect to survive 1.5 million journeys if each trip was 1,000 km, which is the safety level offered to airline passengers.

Multiple car journeys of 1,000 km may sound irrelevant, but the statistics could mean something to the traveller who is considering driving from, say, London to the Côte d'Azur by car: if the purpose is to enjoy the countryside and the local cuisine en route, then drive; if it is to avoid flying for perceived safety reasons, your mathematics are flawed; if you are driving because of an irrational fear of flying, then enjoy the route and good luck - in relative terms you will need it.

The world airline safety average, however, is a very rough guide indeed because of enormous regional variations. Actual safety depends heavily on what nationality the airline is, whether the flight is domestic or international, where the flight is taking place, whether the aircraft is jet- or propeller-powered, and what the prevailing weather is like at take-off and landing.

The world's most statistically safe flight would be with an Australian airline, on an international flight to an American destination in summer (American summer), using a jet aircraft. More about regional variations later. Conversely, the least safe would be a domestic flight in a country with a 'Third- World' economy (specific details later) in a propeller-driven aeroplane (particularly if the propeller is driven by a piston engine rather than a turbine), in bad weather.

Air travellers at the planning stage sometimes ask whether there is an airline safety league table. Surely, they say, the safe airlines will publicise their achievement, proudly laying claim to their place in the league? In fact, even the safest carriers do not dare to. Airline fatal accidents are so statistically rare that even a single fatal disaster could make the top-of-the-league carrier disappear from the top 50 - and what might that do to the clientele's loyalty? Beside which, the airlines know that high places in league tables do not eliminate the basic fear of flying anyway. All a league table would do is to imply that air travel is dangerous. Do travellers require a league table of railway operators? Of coach companies? Of taxi drivers? How would a league table be drawn up? Should it take into account accidents since flying began?... since jets took over?... during the last tenor 20 years? Should the accidents taken into account be those in which someone died, or in which everyone on board died, or include also those incidents in which people were injured? And where does the league table put a brand-new airline? It is unproven, inexperienced, but has not had an accident yet, it could be at the top of the league.

These difficulties of definition are among the reasons why airlines themselves steer clear of selling safety. But above all, selling safety clearly implies that there is something to worry about in the first place.

Probably the best indicator of the safety of any form of travel - if it were possible to get the information - is the size of the operator's insurance premium. If someone has offered you a lift in a car and you want to know how safe a driver he is, ask how much he pays for his motor insurance. The higher it is, the more likely you are to die. Airlines are the same. It is the plain truth that Third World airlines and carriers from developing economies generally pay the highest premiums. The Third World airline market does not, it is true, have the same bargaining power with the insurance underwriters that, for example, the US airlines have. But in the end, it is simply accident rates which determine the rate of the premiums. In the USA, airlines will face annual premiums less than 0.5 per cent of the value of their aeroplanes, whereas some carriers from Africa and South America will pay more than three per cent.

Airlines can be crudely graded for air safety by the continent in which they are based: North American airlines, as a whole, are the most consistently safe; the Middle East has had an excellent record for a long time now; western European airlines offer a high level of safety; Asia, the Indian subcontinent and South-East Asia has a mixture of adequate and bad, with patches of good; Central America is poor, but South America is making an apparently successful attempt to recover from a bad record in the 1980s and earlier; and finally African airlines score lowest for safety, with a few exceptionally good airlines among the bad records.

The disparity is enormous: an African airline is more than ten times as likely to have an accident involving fatalities than a North American, Middle Eastern or Western European one. As for where the accident is most likely to happen, the continents are ranged in a similar order, but the disparity widens still further, with Africa topping the league by far. Finally the majority of accidents happen to domestic airlines - international carriers have a better record on average.

As for the exceptional nations, Australia is the safest, along with the USA, then come the Western Europe and the Middle Eastern nations, and competing for bottom marks are Nigeria, Taiwan, Indonesia and Korea.

The safest airline in the world is Australia's Qantas, which has not harmed a soul since the days of wood-and-fabric biplanes in 1937, when it was known by its original name, Queensland and Northern Territories Air Services. But just to show how misleading - even unfair - an airline league table could be, Qantas, with its half-century perfect record, would not be at the top of a ten-year table chart because it is a relatively small airline. Bigger US or European carriers that had a clear record during the last ten years (even though they might have had a fatal accident in the preceding decade) would be higher in the league table than Qantas because they would have operated more accident-free flights in the period under review.

In December 1990, the US magazine Newsday carried out an airline safety survey of 140 carriers between 1969 and 1990 using some unusual premises in its calculations. Nevertheless the results again confirmed the well-established truth that the airlines of the world's richer nations tend to have the best records.

In that same period, to take a random list, the following international airlines had not had any fatal accidents: Qantas, Ansett (Australia), Aer Lingus, Austrian Airlines, Air Madagascar, Air UK, Braathens (Norway), Cathay Pacific, Finnair, Malaysian Airlines, Sabena (Belgium) and Singapore International.

The biggest safety improvement in aviation's history came with the introduction of jets and turbo-prop engines because the turbines that form the core of both engine types are far more reliable than piston engines. So safety climbed steadily during the late 1950s and in the 1960s as piston-power gradually left the scene. Strangely, there was another upward hike from the 1970s to the 1980s, the reason for which was less clear. But during the last tenyears of the millennium, flight safety, having reached a high level, is improving only slowly. To a large extent it is the law of diminishing returns.

The industry itself is becoming more concerned with 'human factors'. 'Pilot error' has always played a part in some two-thirds of all serious accidents, so now that aircraft technology has become progressively more refined and less likely to fail with disastrous results, the experts are looking for ways of making pilots safer. Aviation psychologists are studying pilot behaviour on the flight deck, communication between pilots, and the way they handle today's computerised cockpits.

There is some concern that aircrews are beginning to feel superfluous in machines that do all their tactical thinking and flying for them. The pilots' attitude to the task has to be totally different from what it once was: formerly the job was to fly the aeroplane, now it is to manage the flight in a progressively more complex and crowded environment. The British Civil Aviation Authority leads the world in the 'human factors' field, demanding of pilots that they take an examination in task-related behavioural psychology as a part of their commercial pilot's licence-qualifying procedure. The intention is that they become more aware of the kinds of human mistakes their environment can lead them to make.

Obviously, there is a search for the reasons why airlines from economically poorer, less sophisticated nations have the worst safety records. There is good evidence that they are more likely to cut corners on maintenance and safety regulations than airlines from richer nations - often because government supervision of standards is less stringent. But the accidents themselves are, as in the richer nations, more often caused by pilot error than by technical failure.

Given the higher Third World accident rates, the implication is that training is less good, or that the pilots' attitude towards their job is different, or both. In the end, psychologists have concluded it is largely a cultural matter.

What is it about the Australian culture that makes its airlines so safe? First, discipline is accepted as the basis of cockpit behaviour. Also authority, while respected by Australians, is not put on a pedestal by them - meaning in this context that if the captain makes a mistake the co-pilot will challenge him. There have been many serious accidents in airline history that could have been prevented if the pilot had challenged the captain's actions. For example, the Japanese are a disciplined race and meticulous in their attention to technical detail; but culturally it is difficult for a subordinate to challenge authority and this cost Japan Air Lines a fatal accident in 1982. Recently in the Asia-Pacific countries the realisation has been taking hold that this culture may have a valid place in society, but not on the flight deck, and training programmes now reflect the need of the co-pilot to be a part of a team rather than just a subordinate. Since 1985 JAL has had a first-class safety record.

When considering 'below-average' airlines as a mode of transport in their home countries, the alternative surface transport should be approached critically, too. In a country where a cash-strapped economy and a laissez-faire culture lets an airline's standards drop, perhaps the same is true of the infrastructure that is supposed to preserve national road and rail safety. It may be true that the national air transport system, while it does not compare well with American or European airline safety standards, is still a relatively safe form of transport in absolute terms.

Finally, airline and airport security has become very much a part of air travel worldwide. In some parts of the world, security is token, but that is often because the perceived risk is low. Lockerbie jolted the airline world into a realisation that the subject of airborne terrorism was a serious one, and airlines and countries that are at risk usually have an adequate security system.

Hijacking is relatively rare now, but it tends to go in cycles. It will come back again. This danger is shown in the case of the December 1999 hijacking of an Indian Airlines aircraft for political reasons. It ended up in Afghanistan where the various national authorities gave in to the terrorist hijackers, releasing them with some of the imprisoned colleagues for the sake of whose freedom they had carried out the hijack in the first place. This kind of capitulation in the face of terror is understandable but flies in the face of all the world's accumulated wisdom on the ways of handling hijack. Sure enough, within a month, some Afghans hijacked one of their own national airline's aircraft to Britain. Capitulation was not forthcoming in the UK and the hijackers have been brought to justice, but the importance of international standards for the handling of hijacking has been suitably underlined.

Meanwhile most hijackings today are not of the terrorist type. Usually they are carried out by unbalanced individuals, or people looking for escape or political asylum. They almost invariably fail.The only workable advice to passengers afraid of the terrorist hijack threat is to decide which airlines are the targets of active terrorist groups, and then to travel with airlines which are not. However, the passengers who take that choice should bear in mind that if they cause the threatened airline's business visibly to suffer they have handed the terrorist a partial victory, encouraging further terrorism.