Problems with ESP from g98 rev. A9

Dear CCLers,
 I'm trying to compare electrostatic potential (ESP) for water at
 different points calculated from a cube file and with non-standard input
 with g98.  However I've noticed that the maximum value for ESP for the
 same geometry does not agree!.
 My input geometry looks like this:
 8       0.00000000  0.00000000  0.00000000
 1       0.67956100  0.87766500  0.00000000
 1       0.67956100 -0.87766500  0.00000000
 If I calculate the ESP in a medium cube, the maximum value is 136.656 at
 (0.015492,0.015492,0.015492).  However, if I use a non-standard input
 and specify that same point I get a value of 136.597912.
 The nonstd input looks like this:
 #P nonstd
  1/38=1/1;
  2/15=1,17=6,18=5/2;
  3/5=1,6=6,7=1,8=2,11=2,25=1,30=1/1,2,3;
  4//1;
  5/5=2,32=1,38=4,42=-5/2;
  6/7=3,15=1,28=1,/1,2;
  99/5=1,9=1/99;
 In both cases I'm using b3lyp/6-31G* with no symmetry for the
 calculations.
 Has anyone seen something like this before?, Which value should I
 trust?, (I'm leaning towards the second one but I'm not sure).
 Thanks in advance,
 Andres
 --
 G. Andres Cisneros, Ph.D.
 NIEHS, 111 T.W. Alexander Dr.
 MD F0-08, PO Box 12233
 RTP, NC 27707
 tel (919) 541 4663
 fax (919) 541 0779