Fashionable opinion has long held that cannabis is a soft drug with few risks to health. Its use has become so commonplace that even those who have not tried it usually have friends and relatives who have done so without ill effects. Such widespread personal experience did much to drive the successful campaign to downgrade it from a class B drug to class C.

There is, however, significant scientific evidence that cannabis is not always benign. A study from the Karolinska Institute in Sweden this week found that when rats were exposed to THC — its main psychoactive ingredient — during the equivalent of adolescence, their brains became more sensitive to heroin. In subsequent experiments, animals with experience of THC took much more heroin.

The findings lend biological support to the “gateway theory” of drug abuse, that early cannabis use makes people more susceptible to heroin addiction later in life. Social factors, and perhaps a genetic propensity to risk-taking, probably also explain why most heroin users have experimented with cannabis first. But the notion that neurological changes in the brain are also important cannot be dismissed.

This is far from the only way in which cannabis can be harmful. It is estimated that smoking three joints a day carries the same risk of cancer as 20 cigarettes. More serious still is a link to mental illness: schizophrenia, bipolar disorder and psychosis.

Scientists have built a compelling case that cannabis can trigger or worsen these psychiatric disorders. While most people’s mental health will be unaffected by the drug, for some it can have catastrophic consequences. Robin Murray, of the Institute of Psychiatry, estimates that as many as 10 per cent of schizophrenia diagnoses can be attributed to cannabis. The risk may be partly genetic; research by Avashalom Caspi, a colleague of Murray’s, has found that a gene variant carried by one person in four multiplies the risk of cannabis- induced psychosis fivefold.

It is no longer possible to contend that cannabis is a safe or mild drug. Critics of the Government’s liberal stance are increasingly claiming that the science is on their side and they are right that it gives cause for concern. It does not necessarily follow, though, that ministers were wrong to reclassify cannabis. Science can offer valuable guidance, which should always set the baseline for policy decisions. But though it has a critical place in any sensible debate, it can contribute only so much.

Science has no view on the most appropriate use of police time and resources, the chief reason that was advanced for reclassification. Medical evidence also reveals nothing about the legal strategies that best dissuade young people from taking cannabis. As Murray says, few teenagers know whether the drug is classified as class B or C, and fewer care. And there is a reasonable if unproven argument that legalisation for adults might be a better way of keeping the drug out of young hands than giving criminals a monopoly on its sale.

Such a policy might make cannabis less attractive to dealers who would sell to teenagers, or send a dangerous message that it is safe, but medical research cannot say which. Evidence suggests that the young, with still developing brains, are most at risk. That, however, could support two different approaches: tough age restrictions and education campaigns about its dangers, or a blanket ban designed to keep it off the streets.

The choice is ultimately a matter of politics. It is essential that rigorous, up-to-date research be taken into account when formulating drug policy, so that risks are considered appropriately. Risks, however, do not automatically require regulation: they must be weighed against the costs and benefits of the measures proposed to control them. Science can inform, but it cannot always decide.

Mark Henderson is Science Editor of The Times