This article discusses theoretical issues relating to an apparent terminological inconsistency between two recent studies involving relational responding. These studies employed a functionally similar protocol to establish contextual cues for arbitrarily applicable relational responding by using a nonarbitrary relational responding procedure; however, one employed the term nonarbitrary regarding this procedure, and the other used arbitrary. Both can be legitimately described as correct, but they use apparently contradictory descriptions because they focus on different aspects of the protocol; in one, the label is based on traditional conditional discrimination task nomenclature, whereas in the other, it is based on the type of relational responding being performed. The current article describes and then explains the issue. In doing so, it touches on an important topic concerning the relation between relational responding and the conditional discrimination procedure.