Numerous primary studies and a recent meta-analytic confirmatory factor analysis (Meta-CFA; Said-Metwaly, Fernández-Castilla, Kyndt, & Van den Noortgate, 2018) have shown that Torrance Tests of Creative Thinking -- Figural (TTCT-F) consists of two factors. However, recent research has raised questions regarding factor analysis of the TTCT-F due to poor discriminant validity and artificial clustering of indices. Additionally, the original meta-analysis did not include databases of dissertations and theses. In this study, we replicated and extended the meta-CFA study of the TTCT-F using 56 correlation matrices from 45 studies and a total of 11,142 participants. We investigated whether the problem of discriminant validity, participants' age, the use of Form A or B, and the use of standardized or raw scores influence the factor structure. Our results show that a two-factor structure remains stronger than a one-factor structure in both replication and extension studies, and this conclusion is robust to score type, participant age, and issues related to discriminant validity. However, further research is needed to determine if this structure applies to less commonly used Form B, which had different factor loadings from Form A. We discuss the implications of our results for creativity research, the psychometrics of TTCT-F, and gifted identification.