Many acquisition studies indicate that across languages, children overgenerate definite articles in indefinite contexts. However, proportions and ages at which children make this error vary, and so do theoretical accounts. Attempting to resolve some of the mixed results, we combined the methods of two different studies (Schaeffer & Matthewson 2005 (SM) and van Hout, Harrigan & de Villiers 2010 (HHV)) and administered them to one group of 82 Dutch-acquiring children aged 2–9 and adult controls (N = 23).1 The results show that definite article overuse takes place in (a) only the youngest age group (2;1–3;7) in the relevant SM indefinite condition, (b) only the two oldest child groups (6;0–9;4) in the HHV indefinite condition, and (c) adults score at ceiling in the SM conditions, while only around 70% correct in the HHV conditions. We argue that (a) the indefinite conditions of the two article choice experiments test different types of knowledge, and therefore their results cannot be compared, (b) the HHV task has more methodological drawbacks than the SM task, rendering its results difficult to interpret, and (c) the results provide less evidence for HHV’s unranked-constraint hypothesis than for SM’s lack-of-Concept-of-Non-Shared-Assumptions hypothesis.