One of the main types of blood cancer is not one but 11 distinct diseases, detailed genetic analysis suggests.
A study in the New England Journal of Medicine found genetic differences explain why some patients respond much better to treatment than others.
The researchers say their findings should help with the development of clinical trials.
Cancer Research UK says this type of study offers new insights into cancer.
The study focused on Acute Myeloid Leukaemia (AML) - there are around 3,000 new cases every year in the UK.
The disease is often aggressive, particularly in older patients - overall survival after five years is about 20%.
Treatment is predominantly chemotherapy and stem-cell transplantation but for many patients it can be difficult to tell how they will respond.
Clinicians currently rely on checks for chromosomal abnormalities and analysis under a microscope. In this study - involving more than 1,500 patients - researchers carried out a far more detailed genetic analysis of the cancer.
They looked at more than 100 genes known to cause leukaemia, and investigated how they interacted.
They found the patients divided into at least 11 major groups, each with their own set of genetic changes and clinical features.
Dr Peter Campbell from the Wellcome Trust Sanger Institute, who co-led the research, said the findings would help doctors to make "very meaningful predictions" about what will happen to patients.
"I could have two patients who had what looked like the same leukaemia under the microscope and I could treat them with exactly the same therapy.
"One of those patients would be cured and one would relapse and die very quickly. What we can see in this data-set is that that clinical variability is strongly predicted by the underlying genetics."
Dr Campbell said he hoped this technique would "filter into clinics" over the next few years as more centres develop their diagnostic resources.
He said a range of exciting new targeted treatments was coming online for some of these genetic changes. Understanding the structure of the leukaemia, he said, would help scientists to develop trials which would bring those drugs to the right subsets of patients.
Dr Elli Papaemmanuil, who co-authored the study, said the findings shed new light on the fundamental causes of AML.
"For the first time we untangled the genetic complexity seen in most AML cancer genomes into distinct evolutionary paths that lead to AML.
"By understanding these paths we can help develop more appropriate treatments for individual patients with AML."
Dr Áine McCarthy from Cancer Research UK welcomed the findings.
She said: "Science such as this continues to offer new insights into cancer which can help us achieve our goal of beating the disease.
"We need to learn more from clinical trials to find out whether tailoring treatment based on these subgroups boosts the number of people surviving the disease."