Using MP2010 UK, If I use teritory manager to display a postcode teritory boundry the information displayed is wrong. there is a significant plotting error-at first sight it looks like it may be that the data is in OSGB and is not being converted to WGS84 before plotting.
I am displaying the map at 1600*1200 screen resolution.
Example: I have turned on the boundry of postcode district RM8
The boundry is wrongly shown as including part of Rosslyn Avenue. clicking a point on the road outside of the drawn area reveals an address (e.g. 65 Rosslyn Avenue, Dagenham, RM8 1) which should be inside the area.
In addition to the boundry marked by a solid line there is a boundry marked by a dotted line. The two lie close, but not exactly, on top of each other, and both are wrong.

Sounds a bit odd - I guess it is possible that one of the datasets was imported without the required coordinate transformation. It would take some detective work to know for sure - probably including ground measurements or against a known "truth" (eg. a 1:10,000 OS map).

MapPoint 2002 did have an error that meant it failed to include the geoid transformation when plotting OSGB coordinates. I was able to show it using known coordinates measured from a 1:10,000 paper OS map, and also to show it was fixed in MapPoint 2004.

Airy36/WGS84 geoid errors in the UK are typically 100m or so - it sounds like the errors you are reporting are much smaller??

100m seems about right: postcode boundries are naturaly 'spikey' as they have to wiggle between the ends of roads: a 100m error can easily cause a spike to wrongly intersect a road.
displaying the map at different screen resolutions does not fix it.
the errors are quite obvious if you zoom in on a displayed postcode boundry!

Sounds a bit odd - I guess it is possible that one of the datasets was imported without the required coordinate transformation. It would take some detective work to know for sure - probably including ground measurements or against a known "truth" (eg. a 1:10,000 OS map).

MapPoint 2002 did have an error that meant it failed to include the geoid transformation when plotting OSGB coordinates. I was able to show it using known coordinates measured from a 1:10,000 paper OS map, and also to show it was fixed in MapPoint 2004.

Airy36/WGS84 geoid errors in the UK are typically 100m or so - it sounds like the errors you are reporting are much smaller??

mappoints data has not come from the same source as an OS paper map... although the boundries would (if plotted correctly) include the correct addresses they are not the same shape.. e.g mp might have a square corner where OS has a curve etc. It seems bizzare that the data could originate from 2 different sources!

Right, but the paper map can be used to check for geoid errors - I did this successully back with MapPoint 2002,2004. I was not working with postcode boundaries which have their own problems (see TMRsimon's reply) but with known surveying points and intersections (a road bridge over a single track rail line in Snowdonia was one of the chosen points).