I was a little bit confused why you had to edit the scale then. When you export to KML, all positions are absolute 'Latitude, Longitude'. I read your previous thread and I see that you are changing the scale BEFORE transforming to Geodetic WGS84. So I guess the issue was in your original map scale information. That makes more sense now. Did you try calculating the scale as I suggested a while back (on march 16, point #2)? It is possible that the distances given in Google Earth are approximated as well since the data is not projected as such. Maybe check their documentation. I think there is an issue with the source coordinate system chosen for the file and that we still have to determine which is the correct one. If we consider that it is a pseudo-mercator (as Mr Barth did), then enter the calculated scale (around 1/13000, I checked it) and enter the anchor coordinates obtained in Google Earth and finally transform the map by performing a coordinate system transformation to the coordinate system 'Geodetic >, World >, WGS84' then the result is not correct when it's imported in Google Earth. To have a (almost) correct result we have to modify the scale before performing the transformation, choosing a value around 1/21000 instead of the calculated value with Google Earth and Illustrator/MP (as Mr barth previously explained we get 1/13000 after the transformation). So if we do change the scale then the data are correctly exported and fit pretty well in Google Earth. But it's not intellectually satisfying ,) and painful when you need to reproduce it many times for several different maps. Again I guess the problem is with the source coordinate system but may be I am wrong. I am still wondering what I missed... if I did. Could you also include some explanation on how you determine the scale? I mean where do you get the real world measure of distance to compare with the distance on paper. A few screenshots are always welcome for that... I think the problem might be in the 'real world' measure of the distance personaly, as I told Lionel. You have to be careful what the measure tool in Google Earth is giving you, it could be approximated. Although it's closed to impossible to guess a projection of map, there's some small signs you can look for to determine what are some of the map projection properties In your case, a big clue is that if you measure distances in different direction (N-S or E-W or diagonally), you always find the same scale ratio with your map. That means that the projection of the map is somewhat equidistant. If the map was in WGS84 of in Pseudo-Mercator, you wouldn't get the same ratio for the scale. It wouldn't be extremely different because it is a large scale map you have, but still here I always got about 1:12,800 scale. So I tried assigning the coordinate system British National Grid (based on Transverse Mercator) - Transverse Mercator is not quite equidistant everywhere, but London is close enough to the central meridian of that coordinate system to be good enough. And with that one, I got some good match in Google Earth. It could be that the actual projection of the map is different, but it is hard to guess further than that. Since that the 'common' coordinate system usually used for British maps, it is the safest bet I can make. Feel free to try some other if you wish. We have the problem of defining the position of a reference point in the British National Grid units (we can easily find the lat/long of a point then we have to transform this lat/long in British National Grid units). We have the same kind of problem on maps of Barcelona, Marrakech, ... which will not be on a British National Grid geodesic system, but they fit on the google map layer we create under our illustrator map. Source.