The study looked at 25 trials over 10 years
The benefits of some cancer drugs may be exaggerated as a rising number of trials are stopped early, experts say.
Italian researchers analysed 25 trials, including some for the breast cancer therapy Herceptin, that were stopped early between 1997 and 2007.
The Mario Negri Institute team said data from many of the recent cases had been used to get drug licences before the long-term impacts were known.
But drug firms said finishing trials early saved lives.
The Annals of Oncology report showed that of the 25 trials randomly chosen, 14 had been stopped in the past three years.
And of these, 11 were used to support applications for marketing authorisation from regulators.
Lead researcher Dr Giovanni Aplone said the increase in early conclusions to trials suggested drug firms were using good interim results to get their products to market more quickly.
But he warned: "Data on effectiveness and potential side-effects can be missed by stopping a trial early."
He admitted there was no hard evidence of this, but said there was an in-built bias in the system because trials were often only stopped early because the results were positive, when this could just be a "random high".
Meanwhile, those that did not show such positive results were given more time to prove their worth.
The team found that the average study duration was 30 months - when the long-term impact could only be judged over years.
The report also said some trials only enrolled less than 40% of the total patients planned.
Researchers said regulators needed to take into account the impact of stopping a trial early when making decisions about licences.
And they added there needed to be more use of independent monitoring committees to verify trial data. Only the largest trials tend to take this approach.
Professor Stuart Pocock, an expert in medical statistics from the London School of Hygiene and Tropical Medicine, agreed the issue was a problem not just for cancer drugs but all kinds of treatment.
He acknowledged trial organisers faced a dilemma when results were positive because those patients involved in the studies, but not receiving the therapies, could lose out.
But he added: "We need proof beyond reasonable doubt to stop a trial early."
A spokesman for the Association of the British Pharmaceutical Industry pointed out that the study looked at only a small number of trials.
He added: "When a trial is stopped early the data is carefully scrutinised and we are confident it is robust.
"Patient safety is our top priority and when results are positive it can cost lives by not getting the product to market."