I will give the valuation theoretical content of local uniformization, which is the local form of resolution of singularities (with respect to a given place of the function field). Zariski proved in 1940 that local uniformization always holds in characteristic 0. But like resolution of singularities, local uniformization is still an open problem in positive characteristic. I will show that this problem is related to the defect, a valuation theoretical phenomenon that appears only in positive characteristic. I will give examples for the defect and discuss two theorems that help to tackle the defect. These theorems lead to two important theorems about local uniformization in arbitrary characteristic: 1) it always holds for so-called Abhyankar places 2) it always holds after a finite extension of the function field (this is a local version of de Jong's result).