Relational reasoning (RR) and divergent thinking (DT) are two critical antecedents of creative problem solving, but the relation between them is not currently well understood psychologically, limiting efforts to support these constructs through education. The threshold hypothesis (TH) is currently the dominant explanation for the relation between RR and DT, and posits that RR fundamentally supports DT, but only up to a point. In this study, quantile regression was used to test the TH among RR and two separate dimensions of DT: originality and fluency. Results generally supported the TH in regards to originality, with RR being significantly positively related to originality, but only in students at or below the median of the originality distribution. However, the TH was not upheld for fluency, which was only significantly predicted by RR at the top (i.e. 9th decile) of the fluency distribution. In general, results suggest that direct instructional intervention of RR strategies may be most supportive of creativity for those students who are simultaneously highly fluent but low-original thinkers.