A link between highly processed foods and cancer has been suggested by French researchers.
They classified foods including cakes, chicken nuggets and mass-produced bread as "ultra-processed".
A study of 105,000 people hinted the more of such foods people ate, the greater their risk of cancer.
A lot of caution is being expressed about the study, but experts said a healthy diet is best.
What counts as ultra-processed
Diet is already known to affect the risk of cancer.
Being overweight is the biggest preventable cause of the disease after smoking and the World Health Organization says processed meat does slightly increase the risk of cancer.
But what about ultra-processed foods?
The team - at Universite Sorbonne Paris Cite - used food surveys on two days to work out what people were eating.
Those on the study, who were mostly middle-aged women, were followed for an average of five years.
The results, in the British Medical Journal, showed that if the proportion of ultra-processed food in the diet increased by 10%, then the number of cancers detected increased by 12%.
During the study:
The researchers concluded: "These results suggest that the rapidly increasing consumption of ultra-processed foods may drive an increasing burden of cancer in the next decades."
But they said the findings need to "be confirmed by other large-scale" studies and research was needed to establish what could be behind the link.
A 'warning signal'
This study is far from the definitive take on ultra-processed foods and cancer.
It cannot say ultra-processed foods are a cause of cancer.
There are also factors that muddy the waters as people who ate a lot of ultra-processed foods had other behaviours that have been linked to cancer.
They were much more likely to smoke, were less active, consumed more calories overall and were more likely to be taking the oral contraceptive.
While the researchers did adjust their analysis for this they say their impact "cannot be entirely excluded".
Prof Linda Bauld, Cancer Research UK's prevention expert, said: "It's already known that eating a lot of these foods can lead to weight gain, and being overweight or obese can also increase your risk of cancer, so it's hard to disentangle the effects of diet and weight."
Overall she said the study was a "warning signal to us to have a healthy diet" but people should not worry about eating a bit of processed food "here and there" as long as they were getting plenty of fruit, vegetables and fibre.
Dr Ian Johnson, from the Quadram Institute in Norwich, said the study had "identified some rather weak associations".
But he criticised the vagueness of the term ultra-processed.
He said: "The problem is that the definition of ultra-processed foods they have used is so broad and poorly defined that it is impossible to decide exactly what, if any, causal connections have been observed."
For Prof Tom Sanders at King's College London, the definition of ultra-processed foods throws up too many quirks.
He said mass-produced bread would be classed as ultra-processed, but a home-made loaf or bread from a posh local bakery would not.
He said: "This classification seems arbitrary and based on the premise that food produced industrially has a different nutritional and chemical composition from that produced in the home or by artisans. This is not the case."
Even the accompanying commentary in the British Medical Journal warned against jumping to conclusions.
Martin Lajous and Adriana Monge from the National Institute of Public Health in Mexico, warned "we are a long way from understanding the full implications of food processing for health and well-being".
They said the study was simply "an initial insight".
