Confidence in fossil ages is a recognized constraint for understanding changes in archaeological and palaeontological records. Poor estimates of age can lead to erroneous inferences—such as timing of species arrival, range expansions and extinctions—preventing robust hypothesis testing of the causes and consequences of past events. Therefore, age reliability must be demonstrated before patterns and mechanisms are inferred. Here we present a generalized quality-rating scheme based on a two-stage set of objective criteria: first, our method assesses the reliability of an age regarding the dating procedure, and second, if the age is based on association, it assesses the confidence in its association with the target vertebrate fossil. We developed this quality rating specifically for Australian applications, but it could be applied to other regions and to longer timescales with some modification. Our method ranks ages in four categories of reliability (A* and A are reliable; B and C are unreliable). In our case study of the late Pleistocene megafauna of Sahul, accounting for reliability (i.e., accepting only reliable ages) reduced the number of useful records within chronologies by 70%; for most species, this greatly affects any inferences regarding the timing and possible drivers of extinction. Our method provides a simple, replicable and general tool for assessing the age quality of dated fossils, as well as provides a guide for selecting useful protocols and samples for dating.