DNA evidence is not exactly reliable in criminal investigation as interpretation of samples can be highly subjective and prone to error, says a new study.

The incredibly small amount of DNA in samples and pressure to gain a conviction can lead to bias, according to an investigation by New Scientist, reports the Daily Mail.

The magazine sent a sample of DNA from a real crime scene to 17 experienced analysts in a US lab. The experts' differing results cast doubts over the technique's reliability.

The sample, from a gang rape, had already been used to convict a man but only one of the 17 scientists came to the same conclusion.

Itiel Dror, a University College London scientist, who helped set up the investigation, said: "It is time DNA analysts accept that under certain conditions, subjectivity may affect their work.'

Christine Funk, a defence lawyer in the US, said: "The difference between prison and freedom rests in the hands of the scientist assigned the case."

The chances of two people having the same DNA fingerprint are between 800,000 and one billion to one. But there are concerns that increasing reliance on tiny samples of blood and saliva, often from more than one person, leaves interpretation open to the scientist's judgment.

"Profiling is generally seen as infallible and it is always able to get its man. But DNA profiling is far from perfect," the New Scientist added.

Kay Francis of the Forensic Science Service, which handles bulk of the police's forensics work, said: "Britain has led the world in terms of breakthroughs in forensic science. North America is quite a bit behind in that aspect."