One negative eigenvalue - No imaginary frequency!!??

Hi CCLers!
 I have got a weird -well, to be honest, a couple of them- output files
 > from g03 where I try to locate the TS of a nucleophilic attack. After
 convergence a frequency job is performed. It results in one negative
 eigenvalue but no imaginary frequency associated to it. Actually it
 seems like there is one imaginary frequency, but gaussian is treating it
 as a traslation/rotation. The lowest (but positive)  mode from the
 vibrational analysis is likely to be the nucleophilic attack but I
 cannot get it to be imaginary.
 I have tried several basis sets and several levels of calculation (HF,
 B3LYP, MP2) with the very same luck. Is there anything I can do to
 locate the "real" TS.
 Also I am enclosing here the very first modes of one of the structures
 in case it helps to understand what is going on:
 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 Full mass-weighted force constant matrix:
  Low frequencies ---  -31.0607    0.0014    0.0024    0.0025   22.5484
 25.9445
  Low frequencies ---   42.3766   64.9775  107.1348
  Diagonal vibrational polarizability:
       624.6067824     849.1987473     446.1087874
  Harmonic frequencies (cm**-1), IR intensities (KM/Mole), Raman
 scattering
  activities (A**4/AMU), depolarization ratios for plane and unpolarized
  incident light, reduced masses (AMU), force constants (mDyne/A),
  and normal coordinates:
                      1                      2                      3
                      A                      A                      A
  Frequencies --    18.6726                63.9044               105.9740
  Red. masses --     5.4210                 4.4793                 3.1978
  Frc consts  --     0.0011                 0.0108                 0.0212
  IR Inten    --    21.2490                13.5758                 6.6021
 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 Thanks a lot!