The next generation of large sky surveys will return panchromatic
images but not spectra.  The photometric redshift estimates from these
surveys are expected to be good enough for precision cosmology.
However, naive use of these noisy distance estimates will lead to
biases in measurements of most galaxy scaling relations, thus
compromising the ability of such surveys to constrain models of how
galaxies formed.  I will discuss a number of methods - convolution,
deconvolution, maximum likelihood and background subtraction - which
eliminate such biases, thus enabling photo-z surveys to provide sharp
constraints on galaxy formation models as well as cosmology.