There can be many good reasons for "disagreement". It might help to ask yourself the following: What approximations are being made in the calculation? What spectra are you comparing to and how was it measured?

Hello,

I agree fully with the answer of Bradley!

please send infos for the answerers about your molecule. What a kind of substituent (s) and purity  for UV-Vis measurements  

Infos about solvent (cut off)  and spectrometer (single, dualbeam or diodes array. What a kind of cuvettes for absorption or later emission?

Concerning Gaussian 09, please read this link:

 http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3107493/

Send please spectra as pdf files  (lambda max) and the Lambda max values  of the  Gaussian output. (as pdf files)

(see table 3 in the joined link)

JRG

The computed TD-DFT optical spectra are often shifted in energy with respect to experimental ones.

It depends mainly on the specific type of XC functional adopted.

In general, I would expect that functionals with significant HF exchange (such as B3LYP, PBE0, M062X, M06HF...) were to yield UV-Vis spectra somewhat blueshifted with respect to the experimental ones.

Thank you Mr Bradley, Mr. Jean and Mr. Francesco for answering my question. 

This is my input,

%Chk=Klorofil_b.chk

# B3LYP/cc-pVDZ TD=NStates=6 SCRF=(Solvent=DiethylEther)

I compare my result with this, http://www.ncbi.nlm.nih.gov/books/NBK22535/

I using this my ipnut, but some professor told there is a change in molecule structure, like something must be modified. I don't know what he mean about it, do you have any idea ?

Thank you

Without knowing the details there is little to say. Perhaps there is aggregation. Perhaps some explicit solvation needs to be taken into account. Perhaps the basis set is too small. Why should B3LYP even give the correct answer? There are known problems with B3LYP for excitations leading to errors from the 'exact' theoretically correct values. The error magnitude mentioned above could be from any of multiple causes.