This article reconsiders one of the classic articles in the second language literature, Schachter's 'An error in error analysis', in the light of subsequent work on avoidance. Such work is essentially of two kinds, either building on and refining the concept of avoidance, or offering alternative explanations for Schachter's findings. While avoidance as a genuine phenomenon is not proved or disproved by Schachter's data, hypotheses based on her figures suffer from the lack of methodological detail in her original study. In order to be able to establish whether avoidance is a feasible explanation for relative underproduction by a group of learners, it is necessary to look at the first language form, distribution and function of the entity supposedly being avoided in the L2 as well as the means being used to establish whether and to what extent the entity is already part of the L2 knowledge of members of that group.