We present results of an investigation of clustering evolution of field galaxies between a redshift of z \sim 1 and the present epoch .
The current analysis relies on a sample of \sim 3600 galaxies from the C alar A lto D eep I maging S urvey ( CADIS ) .
The multicolor classification and redshift determination is reliable up to I = 23 ^ { mag } .
The redshift distribution extends to z \sim 1.1 , the resolution is \Delta cz = 12000 km s ^ { -1 } .
Thus the amplitude of the three-dimensional correlation function has to be estimated by means of the projected correlation function w ( r _ { p } ) .
The validity of the deprojection was tested on the Las Campanas Redshift Survey ( LCRS ) .
The LCRS also serves as ” local ” measurement .
We invented a new method to overcome the influence of redshift errors on w ( r _ { p } ) .
For evolution of the clustering strength the ansatz \xi ( r _ { com } ,z ) \propto ( 1 + z ) ^ { q } is used .
For the galaxies as a whole the evolution parameter turns out to be q \approx - 1.9 , according to the prediction of linear theory .
A formal dependency on the cosmology is presumably due to the small number of fields observed .
However , the measured clustering growth clearly depends on Hubble type .
At z \sim 1 early type galaxies are already much stronger clustered , an increase with q \simeq - 1 is sufficient to explain the present day amplitude of the correlation function .