Assuming, that it's Sql Server table 'testun' with one field 'un int'. :)

SELECT t1.un+1 AS uns, 
		(SELECT MIN(t3.un-1) FROM testun t3 WHERE t3.un-1 >= t1.un+1) AS une
	FROM testun t1
	WHERE NOT EXISTS (SELECT * FROM testun t2 WHERE t2.un = (t1.un+1))

Here's solution for VFP

SELECT t1.un+1 AS unp, t1.un-1 AS unm, t1.un ;
	FROM testun t1 ;
	ORDER BY t1.un ;
	INTO CURSOR crs1

SELECT t1.unp AS uns, t1.un, RECNO() AS rn ;
	FROM crs1 t1 ;
	WHERE NOT EXISTS (SELECT * FROM crs1 t2 WHERE t2.un = t1.unp) ;
	INTO CURSOR crs2

SELECT t1.unm AS une, t1.un, RECNO()-1 AS rn ;
	FROM crs1 t1 ;
	WHERE NOT EXISTS (SELECT * FROM crs1 t2 WHERE t2.un = t1.unm) ;
	INTO CURSOR crs3

SELECT t2.uns, t3.une ;
	FROM crs2 t2 ;
		JOIN crs3 t3 ON t2.rn = t3.rn ;
	INTO CURSOR crsResult

>Hi everybody,
>
>Ok, I was inspired by Peter. Here is a very simple problem, but neverless, let's find out the most elegant solution.
>
>There is a table with some sequence unique numbers:
>
>1
>3
>4
>8
>15
>...
>
>No pattern. Your goal would be to find out all missing numbers (of course, only whole numbers).
>
>For more advanced, I would make the task more difficult: find out all missing simple (? oops, what's the correct math term?) numbers. The latest problem seems complicated and I don't know solution myself yet...