In this talk I will show how we can study cosmolgy in a photometric redshift galaxy survey, by means of the angular clustering of galaxies. Previously to fit your data to a cosmological model, the need for a representative, clean and reliable galaxy catalog imposes many constrains in the selection of your data, from the day the data was taken, up to the final galaxy catalog used in the cosmological Analysis. I will try to introduce those issues that are most important for the analysis of galaxy clustering: data reduction and detection limit, catalog pruning, sample selection, photometric redshifts, star/galaxy separation and the need for a detailed angular and depth mask. Once a neat catalog is build upon the raw data of the survey, the cosmological analysis can start confidently. In this context, I will show as an example, the last cosmological results obtained from the DR8 SDSS-III photometric sample, conveniently corrected from systematic errors.