The decision to end key stage testing in Wales is supported by the findings of research on how best to assess the outcomes of learning.
After nearly three years, the "assessment systems for the future" project concluded that, while no approach is without problems, the use of teachers' judgements has far fewer disadvantages than a system based on external tests.
A good deal of assessment of learning is, and always has been, conducted by teachers - for keeping a record of progress, reporting to parents, and guidance about course options. But, when it comes to assessment for uses outside the school, there is often a preference for tests and examinations.
This is despite evidence of their shortcomings: that they assess a rather narrow selection of learning outcomes; are of limited reliability; have a narrowing impact on the curriculum and teaching methods; and make heavy demands on schools' resources in terms of time and money.
To avoid the new system in Wales having some of the same failings, it is essential that there is greater consistency between what is valued, what is taught, and the learning that is assessed and reported.
In theory, teachers can build up a picture of pupils' achievement across the full range of activities and goals as part of their regular work. But what if evidence is missing or not recognised? What about the reliability of teachers' judgements and the impact on workload?
Facing such issues is important if the benefits of ending testing are to be realised. Messages from those working in systems where only teachers' assessment is regularly used underline the need for professional development, moderation, and time for a new system to bed-in before judging whether it works.
There are clear implications here for those responsible for education policy. A generation of teachers, particularly in primary schools, has been dependent on tests and will need time to develop new practices. Initial and professional development courses have to prepare teachers for their roles in assessment far more thoroughly than in the past.
There is evidence that teachers' judgements can be inaccurate and biased.
But this comes from comparing test results with assessments by teachers who have had no guidance or training in the process.
Other evidence shows that when criteria are well specified and understood, teachers are able to make judgements of acceptable reliability.
Tests may be considered more reliable, but in fact they are inherently limited because tests can only sample the full range of learning outcomes.
In KS2 tests, it is estimated that around 30 per cent of pupils will be given the wrong level. Some form of moderation is usually used to increase the reliability of teachers' judgements.
This most usefully takes the form of teachers of the same subject or age group meeting together, discussing the profiles of particular pupils and aligning their judgements of best fit with the reporting level descriptions.
Such meetings are in themselves valuable professional development experiences, leading to better understanding of curriculum goals and how to achieve them.
When results are to be used outside, meetings across schools are needed.
But again these have benefits for teaching and learning as well as for good assessment.
Although there is already experience of within-school moderation, particularly for the lower age groups, any extension of this is bound to raise objections about the time involved.
In terms of teachers' time in their own class, the evidence is that both teaching and learning time can be saved by making more use of teacher assessment.
This is because when there are external tests, teachers spend a good deal of time on test preparation and practice - often at considerable cost in buying commercial tests. If only half of this time is saved, because assessment activities still take up some, there is more time for teaching and learning. However, this does not solve the problem of time for moderation meetings.
Local authorities and school management need to plan the school year and sessions to allow time for teachers to meet when the pupils are not at school (such arrangements have been in place in Scotland for some years).
It requires school management to ensure that everyone understands the reasons for changes. Parents, teachers, pupils and other users of assessment results are more likely to recognise the considerable benefits of moving from a system dominated by tests to one which makes good use of assessment by teachers if there is openness about its processes and issues.
Then users of assessment could see how results have been reached, and so develop trust in teachers judgements. It would also make it possible for pupils to have a part in their own assessment.
Wynne Harlen OBE is visiting professor of education at Bristol universityGor more information see 'The role of teachers in the assessment of learning', published by the Assessment Reform Group, at www.assessment-reform-group.org