A  Finite  Element  (FE)  model  which  can  precisely  predict  the  actual  dynamic  responses  of  a  bridge  is necessary  for  accurate  condition  assessment  and  health  monitoring  purposes.    However,  most  of the bridge FE models  initially developed  for  design purposes may  not  be  able  to  accurately simulate  the  current response of the bridges due to various  reasons.  Some  of  these  reasons  may  be  attributed  to  high  level  of  model  idealization, uncertainties  involved  in  modeling  complex  structural  components  and  changes  in  the  boundary conditions or material properties from the time of construction. Therefore, the initial FE model of the bridge needs to be calibrated. This paper presents the implementation of a sensitivity-based  model  calibration  procedure  to  minimize  the  error  between  the  numerical  and  experimental  modal  properties of the Carquinez  Bridge  (officially  named  the  Alfred  Zampa  Memorial  Bridge);  a  1.06  km long  suspension  bridge  located  in  Vallejo,  California.  Twenty two Narada wireless sensing units were installed on the bridge to measure the ambient-induced vibration responses of the bridge. The modal frequencies and modal shapes were extracted from the measured vibration responses. Sensitivity analyses were  performed  to  identify  the  most  sensitive  structural  parameters  of  the  FE  model.  A multi-variable  sensitivity-based objective  function  was  developed,  and  an  optimization  problem  was  solved  to  update  the  identified structural  parameters  through  an  iterative  procedure.  The  updated  FE  model  of  the  bridge  provides  improved modal  properties  which  closely  match  with those from the measurements. The  calibrated  FE  model  is  expected  to  generate  much  more  accurate  results  for  the  post-earthquake evaluation of the structural performance.