NNLO PDFs for the LHC

We consider some trends, achievements and a series of remaining problems in the precision determination of parton distribution functions. For the description of the scaling violations of the deep-inelastic scattering data, forming the key ingredients to all PDF fits, a solid theoretical framework is of importance. It is provided by the fixed flavor number scheme in describing the heavy-quark contributions which is found in good agreement with the present experimental data in a very wide range of momentum transfers. In this framework also a consistent determination of the heavy-quark masses is possible at high precision. The emerging Drell-Yan data measured at hadron colliders start to play a crucial role in disentangling the quark species, particularly at small and large values of $x$. These new inputs demonstrate a good overall consistency with the earlier constraints on the PDFs coming from fixed-target experiments. No dramatic change is observed in the PDFs in case of a consistent account of the higher-order QCD corrections and when leaving enough flexibility in the PDF shape parameterization.