The threshold hypothesis (TH) assumes the existence of complex relations between creative abilities and intelligence: linear associations below 120 points of IQ and weaker or lack of associations above the threshold. However, diverse results have been obtained over the last six decades--some confirmed the hypothesis and some rejected it. In this paper the threshold hypothesis was examined on a large sample of middle and high school Polish students (N = 921). Intelligence was measured by Raven's Progressive Matrices (RPM) and creative abilities by the Test for Creative Thinking-Drawing Production (TCT-DP). Results were analyzed as raw test scores, Item Response Theory (IRT) scores and Confirmatory Factor Analysis (CFA) scores to examine three thresholds (+0.50 IQ SD, +1 IQ SD, +1.33 IQ SD). It was found that confirmation or rejection of the TH depends strongly on both an analytical strategy and theoretical decisions required for acceptance/rejection of the TH. If significant correlations between intelligence and creative abilities below the threshold and non-significant correlations above the threshold are treated as confirming the TH, such confirmation is observed at 115 points on the IQ scale. However, if confirmation requires higher correlations below the threshold than above it, the TH is less likely to be confirmed. We discuss theoretical and empirical issues which may lead to the conclusion that threshold hypothesis may actually be an artifact of data analysis. (Contains 1 figure and 5 tables.)