Geometry Optimization using CIS and AM1-CI

Dear CCLers,
 I'm studing the geometry of the first excited state using AM1-CI
 (implemented im MOPAC 6) and CIS (implemented in GAUSSIAN 94). I increased
 the size of the CI in AM1-CI and the final geometry changed, as I
 But the optimized geometries using CIS method for diferent
 number of excitations used in the calculation were the same. It's a
 particularity associated to the sistem I'm working or it's a particularity
 of the method?
 Does anybody have references of geometry studies of the excited state
 changing the number of excitations and using the CIS method?
 Thanks in advance,
 Demetrio FIlho
                 Demetrio A. da Silva Filho
                      UNICAMP - IFGW
                    Predio D - Sala 17
                 CEP 13083-970 C.Postal 6165
                   Campinas - SP - Brazil
 "Se nao houver frutos, valeu a beleza das flores. Se nao houver flores,
 valeu a sombra das folhas. Se nao houver folhas, valeu a intencao da
 semente." Henfil