There is no longer any doubt that the safety standards between cargo and passenger operations are massively different - and the latest figures prove it. According to a study by the Netherlands National Aerospace Laboratory (NLR), the most dangerous commercial aviation activity in the world is ad hoc cargo charter flying in Africa (Flight International, 14-20 March), which has a hull loss/fatal accident rate more than 11 times the passenger operator average. But even among major scheduled cargo operators in North America, the NLR study reveals, the accident rate is three times that for their passenger counterparts.

There is now evidence that the problem starts with the national regulators, and that it is time for the International Civil Aviation Organisation (ICAO) to raise the standards of freight operations.

The traditional explanations ventured for the differences are that freighters are, on average, older than their passenger fleets, and they are more often flown close to their maximum takeoff weight limits. Freighters more often depart at night when the aircrew's body clocks are in their "circadian low" periods. If this argument were advanced by the freight airlines, it would rightly be seen as a dreadfully passive attitude, an excuse for not being good at what they do. But if those are the problems freight airlines face, why don't they concentrate on overcoming the problems specific to their trade?

The NLR study, without actually debunking the traditional explanations, introduces an indicator that the differences between passenger and freight operations are over-hyped. Freighters do not usually have different kinds of accidents from their passenger counterparts, the NLR says, they just have them more often.

There are, however, specific dangers which are much greater threats to cargo aeroplanes, and recent experience bears this out. Bad cargo loading, putting the aircraft's centre of gravity outside limits, can happen more easily with main deck freight, and load shift is more likely because there are usually no bulkheads. There has been a catalogue of fatal or serious load-caused accidents in the last few years, and probably many more in those countries where investigative reports never see the light of day.

This is a matter of increasing seriousness, because pure cargo operations are growing as a proportion of total traffic.

If any agency, like the US Federal Aviation Administration, for example, cites a cost-benefit analysis for not upgrading regulations for freight operations to passenger standards because only a crew of two stands to get killed, they have ignored more than half the equation. Freighters have consistently been the greatest killers of people who live near airports, because it is just after take-off or just before landing that a badly loaded freighter goes out of control. The worst example was in January 1996, when a Scibe Airlift Antonov An-32 at Kinshasa, Zaire, could not rotate for take-off, killing 300 people and injuring 250 in its overrun.

The FAA is not the only national agency at fault, but its attitude to freighter safety speaks volumes, providing a poor basis for an airline safety culture: 17 years after traffic alert and collision avoidance systems (TCAS) were mandated for passenger aircraft, they are still not compulsory on freighters.

Even after the ValuJet crash, below-deck cargo compartments in freighters do not have to be fitted with fire extinguishers like their passenger counterparts, despite the fact that freighters carry the most hazardous cargo. The crew do not have escape slides. There is no requirement for airlines to train the pilots, who have to sign the load sheets, to recognise a badly loaded or badly secured cargo when they go into the hold to inspect it. Indeed, cargo management does not feature among all the theory on which pilots are examined to earn their licences.

This failure to train pilots in safe cargo practices and how to recognise danger when they see it seems to be universal. This is where ICAO should come in, because somebody has to start the ball rolling.