The proviso has been the traditional legislative criterion, and its formulation has usually used the concept of a substantial miscarriage of justice. That is, an error at trial would not require the quashing of a conviction if it was not a substantial miscarriage of justice.
There was therefore a distinction between miscarriages of justice that were not substantial, and those that were. The former could be termed harmless errors, and the latter errors that may have affected the result of the trial.
But what does "may have affected" mean for this latter group?
In R v Sarrazin 2011 SCC 54 the majority declined to draw distinctions between various grades of risk of affect on the outcome. At  Binnie J said:
Only one submission to the select committee on the criminal procedure bill mentioned this point.
Section 232 of our Criminal Procedure Act 2011 materially avoids any distinction between substantial miscarriages of justice and those which are not substantial. The appeal court must allow an appeal if it is satisfied that a miscarriage of justice had occurred, and that means that there is a real risk that the outcome of the trial was affected by the error. There are other aspects, but these are the ones directly in point here.
How will an appellate court interpret "real risk"? Is there to be a gradation of risks, of the sort that the Supreme Court of Canada majority found unacceptable? Or is a real risk anything more than a harmless error or one which occurred in any case other than one where the prosecution's case was overwhelming?
As the dissent of Deschamps, Rothstein and Cromwell JJ in Sarrazin illustrates, opinions can differ over whether there was a reasonable possibility that an error could have affected the verdict.
In the circumstances of that appeal, the majority's approach, which was that failure by the judge to put to the jury the alternative of an attempt amounted to a substantial miscarriage of justice, is preferable to the minority's speculation on how the jury was thinking. Under the New Zealand reformed law, the majority's reasoning would lead to a conclusion that the trial had been unfair because it was not according to law. That is an alternative ground for finding a miscarriage of justice under s 232.