Information on reference values is one of the great demands in ecological restoration, as they enable evaluating the restoration progress and taking adaptive management measures. Here we aimed to: (i) verify whether ecological indicators are influenced by the area of reference ecosystems (mature fragments); (ii) determine reference and comparison values for comparing the restoration progress; and (iii) check how long restoration forests take to reach naturalness values that are like the reference ecosystems (and can therefore be considered restored). We used the Brazilian Atlantic Forest as an object of study and compilated information about 967 secondary, mature and restoration forests in a wide geographical extension. In total, 14 ecological indicators were assessed in a sampling area of 1,928,024 m(2). We used simple linear regression to analyze the data and descriptive statistics for creating the reference and comparison values. We found that ecological indicators were not related to the area of mature fragments. Thus, they are useful for monitoring reference ecosystems of different sizes, and more attention can be given to the degree of conservation of the ecosystems. We defined intermediate and final reference values from secondary and mature forests, respectively. We also established comparison values for several restoration stages using data of planting, assisted and unassisted restoration areas. At the same time, we applied the concept of naturalness in restoration and obtained a continuum of naturalness, which was very useful for tracking the restoration progress. With the reference and comparison values determined, restoration practitioners can evaluate their restoration areas by comparing their monitoring results with ours. Moreover, by using the naturalness concept we verified that indicators can have different restoration trajectories. Some indicators reached similar naturalness levels as references from the beginning of restoration, but others required between one and two decades to recover. This study is the first to create evaluation criteria for forest restoration based on a large data set, and we hope that similar studies can be developed in other biomes Ver menos