A new study has found a link between vitamins and certain cancers, surprising even the researchers who carried it out. The latest study found that problems arise when people consume vitamins well in excess of the FDA recommended daily dose. This study is actually one of a long line of papers on this subject—the search for a link between vitamins and cancer goes way back.

There are scores of articles in JSTOR examining links between multivitamins and negative health impacts. The vast majority of those studies, however, never found a problem. The conclusions of a 1997 review article on this topic are typical. Examining research up to that point, the study concluded that multivitamins had either no impact or slightly reduced cancer risk.

Of course, the devil is in the details. Out of hundreds of studies reviewed, some found increased cancer risk, while others found benefits. Results were influenced by pre-existing cancer risk and cancer type, so the generally pro-vitamin conclusions represent the overall on-balance results of all the studies combined (excepting vitamin E, which was consistently associated with decreased cancer risk). In a common refrain for these studies, the authors believed that more research was necessary to sort it all out.

The other issue is appropriate vitamin dosage, and here the research is a little clearer. In most cases, the advice is to take after Goldilocks when it comes to supplements: not too much, not too little. For example, vitamin A is an important supplement, but in high doses it can cause mutations in developing embryos, even serious congenital malformations. At the same time, a different supplement, folic acid, is required in moderate doses during pregnancy to prevent birth defects. Basically, expectant mothers interested in vitamin supplements should read the label carefully.

So what’s the low down? We still don’t know. This most recent study can be added to the large pile of conflicting studies, but it is possible that the overall mild protective effect of multivitamins can be chalked up simply to good nutrition, affluence, lifestyle, etc. It is very hard to tease out a concrete effect from correlation studies, and despite bland assurances it is not really so straightforward to account for all the differences in age, gender, income, pre-existing risk, etc. It’s probably best to just eat as balanced a diet as you can afford.