Dear Jim,

Thank you very much for your help.
I am pretty sure my parameters are correct, because the field map correction in FSL worked fine for the same data. However, I would prefer to integrate the field map correction with the realign-and-unwarp job in SPM. So, in order to exclude problems through potentially wrong settings of parameters in the FieldMap toolbox, I made some comparison with different parameters settings.

1. I am sure about the short/long TE of the field map, so I didn't change them. (We used a typical Siemens sequence with 4.92ms/7.38ms)

2. I also edited the Fieldmap.m file for Siemens data, with the code suggested by Chris Rorden (http://www.mccauslandcenter.sc.edu/CRNL/tools/fieldmap).

3. I tried with and without the option "Mask brain" during creation of the field map. This does not change the anterior shift, apart that non-brain tissue is shifted too, without masking.

4. I tried with and without the option "Match VDM to EPI". This does not change the anterior shift.

5. I tried with +ve/-ve. In case of -ve the shift is into anterior direction, while with +ve the shift is in posterior direction with a comparable magnitude. Also, the distortions in the frontal pole are affected differently. They get less strong with -ve and stronger with +ve. So, I assume that -ve is correct in my case.

6. I tried to calculate the "Total EPI readout time" in different ways:
6.1. Total EPI readout time = echo spacing = 0.52ms; This removed actually the shift, but also the distortion corrections in the frontal pole. Basically the EPI looked the same before and after unwarping.
6.2 Total EPI readout time = echo spacing * number of lines / acceleration factor= 16.64ms; This gives some distortion correction in the frontal pole, but also the previously reported shift in anterior direction
6.3 Total EPI readout time = echo spacing * number of lines = 33.28ms; This roughly doubles both, the amount of distortion corrections and also the shift in anterior direction.

So, assuming that my "Total EPI readout time" was calculated correctly with 16.64ms, none of the other parameters resolved the problem of the anterior shift. Are there any other parameters I could try to manipulate?
I also assume that my basic usage of the FieldMap toolbox was correct, because it worked correctly with the example data from Chris Rorden. However, Chris Rorden's data have the same slice number and voxel dimension for field map and EPI. For this reason, I started to doubt, that the usage of different parameters in the field map might not really work.
I would be very happy about any further suggestions.

Thanks,
Benno