National forest inventories represent a fundamental source of data and knowledge for forestry and environmental policy and allow for the production of national and regional level statistics on forests. The value of these statistics confirms the need for a sampling design that adequately delivers representation by reducing sampling error, but also for a data quality process that limits the non-sampling errors. The article summarizes the quality control procedures of the three sampling phases adopted in the Italian national forest inventory, carried out between 2003 and 2006. The development of an integrated system of actions and controls which are able to limit subjective interpretations, in order to guarantee harmonized information all over the country, was a considerable effort within the overall project. Critical points to be considered were the consistent number of measures and evaluations undertaken during the three inventory phases, the high variability of observed attributes, the consistent number of surveyors involved, and costs of quality control, especially those related to fieldwork. At the end, examples on the overall quality of the classification performed on land cover and vegetation are discussed, as well as the impact of classification errors on the total forest area estimates.