Raw Input		Desired Output
Line 1 AAAA Line 2 BBBB Line 3 CCCC Line 4 DDDD Line 5 EEEE Line 6 FFFF Line 7 GGGG Line 8 HHHH Line 9 IIII		Line 1 AAAA Line 2 BBBB Line 3 CCCC

Script and Comments

Script₁

[ 1] :loop
[ 2] 1,5{
[ 3] $d
[ 4] N
[ 5] b loop
[ 6] }
[ 7] $d
[ 8] N
[ 9] P
[10] D

Comments

'Pattern space' is abbreviated to 'PS'.
Steps [1] thru [6] constitute a loop. When sed exits this loop, PS will contains Line 1 thru 6 of the processed file. For files containing less than 6 lines, they will be discarded by the command 'd' in step [3] .
When Line 6 of the file has been read, unless the last line is reached, we will each time
- Append one next line to PS.
- Print the first line in PS.
- Remove the first line in PS.
When the last line has been read, PS contains all the last 6 lines of the processed file, all we have to do is discarding them via command 'd' in [7].
A refined version is shown in Script₂.

Script₂

[ 1] :loop
[ 2] $d
[ 3] N
[ 4] 2,6b loop
[ 5] P
[ 6] D

Comments

This is a refined version of Script₁.
Be careful! The address in Step [4] is 2,6 rather than 1,5, which is used in Step [2] of Script₁.