Since its founding five years ago only 4 per cent of all its cases have been referred back to the court of appeal. And this low referral rate is a deliberate choice of the CCRC not to challenge the judicial system but to 'second guess' what the High Court will do. A practice much used by the High Court judges, in deciding what juries would think, and condemned by a great majority of leading criminal lawyers and barristers and made illegal by the House of Lords.

MOJUK has always maintained that you can't have the same judges who sent people to prison, hearing their appeals. Experience has shown time and time again that the main function of the Court of Appeal, is to uphold the decisions of previous courts.

Appeals should be heard by an independent court, a court which will not have the restraints of the present system, and it is high time this should be set up.

As Sally Clark recently found to her cost, a verdict reached by the criminal justice system is almost set in stone. Rightly, it is not easy to overturn a "guilty" finding by the courts. But when a mistake has been made, does the appeal process provide an adequate safeguard against a wrongful conviction? Today Sir Frederick Crawford, the chairman of the Criminal Cases Review Commission (CCRC), appears before the Commons Home Affairs Committee.

It is nearly five years since the commission was set up to take over from the Home Office C3 department the job of tackling alleged miscarriages of justice. Its creation was a landmark; public and statutory recognition that the system could get it wrong in the wake of cases such as the Birmingham Six and Guildford Four and a determination to create a body that would operate openly and efficiently.

The commission took a backlog of some 250 cases from the Home Office and Northern Ireland Office. A spokesman said: "There were historic cases such as Hanratty and Derek Bentley and others going back years . . . the files arrived in a pantechnicon." The caseload burgeoned and now the commission receives 800 a cases a year (the Home Office figure was about 700.) Its task is crucial. Bob Woffinden, an investigative journalist who specialises in miscarriages of justice, says: "It remains a near-impossibility to win a case on a first appeal (before the Court of Appeal) - so much so that people feel they must tread water until after that, and then get on the business of going to the CCRC."

To date it has received 4,830 applications, but few will survive through to the Court of Appeal. About a third are thrown out because the person has not been through the appeal process or has no fresh ground for argument. Most, as with the Home Office, are turned down. In all, 161 cases have been to the Court of Appeal in five years, 4 per cent of all cases. The Home Office referral rate was nearer 1.5 per cent.

The CCRC has come under fire for delays and dealing with "dead" cases while prisoners languish behind bars. In its defence, it says the backlog is down from 1,200 to 300 - and the average wait is 14 months for those in custody and 19 months for those at liberty or deceased. And it argues that cases such as Hanratty or Mahmood Mattan, a Cardiff man hanged for murder, justify attention. "These cases have blighted families' lives - to have a person hanged for something they did not do . . . we'd be failing if we did not look at these cases. They are only a small percentage and there has to be a close relative involved."

But a referral to the Court of Appeal is only half the battle. The appeal judges themselves then decide if a conviction should be quashed or a case sent for retrial. The CCRC refers cases to the court if it thinks there is a "real possibility" of the appeal being allowed. So, argues Woffinden, the CCRC is in effect always second-guessing the Court of Appeal. "Even if it thinks there has been an injustice, but does not believe the appeal court will allow it, it won't refer." In a sense, he says, the c. But the CCRC insists that people must have new evidence or legal points not heard at the original trial or appeal. The test may seem stiff, it says, but "we are not here to give everyone a second chance".

The Court of Appeal nonetheless does its own rigorous sifting. Of 161 cases sent so far, 95 have been heard. Two thirds have been quashed and the rest upheld. The percentage has hovered at between two thirds and three quarters, but the past year has seen a rise in rejected appeals, prompting concern, Woffinden says, that the court is taking a tougher line. Lawyers cite cases such as that of Stephen Craven, convicted of murdering a young girl after a fracas in a Newcastle nightclub. His appeal was rejected. Another case was that of Eddie Gilfoyle, convicted of the murder of his wife, Paula, on The Wirral. His lawyer, Campbell Malone, argues that the pendulum in the appeal court has swung back and that convictions are sustained even in the face of compelling new evidence.

What is happening, the lawyers say, is that the court is increasingly considering such evidence itself, in effect doing the job of the jury, then rejecting it. The new test the court now applies is whether a conviction is "unsafe", instead of the old "unsafe and unsatisfactory" or "material irregularity". This gives judges greater leeway to assess the totality of a case.

A recent House of Lords ruling made clear that the Court of Appeal must not put itself in the position of a jury on fresh evidence, Woffinden says. But while that is welcome, other concerns persist: lawyers detect a trend for the Crown, not just the defence, to put forward new evidence - in effect, mounting a fresh prosecution. John Batt, a solicitor friend of Sally Clark who is serving life for the murder of her babies, questions the court's make-up. The judges appear to be chosen at random. Only one judge in Clark's appeal was from the Family Division, he says.

But such judges have much greater experience of child abuse. If the appeals machinery is to work, laywers say, the terms of the CCRC should be widened to refer all cases where there is a real possibility of a miscarriage of justice. The role of the appeal court should be clarified so it does not usurp that of the jury; and where there is cogent new evidence, convictions are quashed outright or sent for a retrial. In that way, juries, not judges, can decide if the system got it wrong.