The r-process involves neutron-rich nuclei far off stability for which no experimental cross sections are known. Therefore, one has to rely on theory. The difficulties in the predictions are briefly addressed. To investigate the impact of altered rates, a comparison of r-process production in hot bubble models with largely varied rates is shown. Due to the (n,gamma)-(gamma,n) equilibrium established at the onset of the r-process, only late-time neutron captures are important which mainly modify the abundances around the third r-process peak.