The factor structure of the Five Facets Mindfulness Questionnaire (FFMQ) seems to vary across samples depending on whether meditators or non-meditators are studied and whether a sample is analyzed before or after mindfulness-based cognitive therapy. The current study illustrates the inconsistencies typically found (e.g., whether all five facets can load on an overall construct of mindfulness), as well as provides and tests alternative explanations in three samples with different levels of meditation experience (i.e., current meditators, past meditators, and non-meditators). Altogether, 2247 German-speaking volunteers completed the FFMQ and reported their meditation experiences online. Results showed that the scaling of three facets of the FFMQ (i.e., observing, describing, and non-judging) were constrained in all samples. The past meditators revealed unique features in terms of their mindfulness level: (1) stopping practicing meditation reduced their levels of mindfulness in facets of awareness, non-judging, and non-reacting, yet observing and describing seemed to remain and (2) those past meditators with intensive trainings scored higher in all five facets than those past meditators who practiced less. The CFA yielded a good fit in all three samples. A hierarchical factor analysis showed how the factors unfolded from level to level and demonstrated that in particular the observing facet loaded on the overall construct of mindfulness differently across the three samples. The empirical results confirmed the alternative interpretations on why the discrepancy regarding the loading of the “observing” facet on an overall mindfulness construct occurs, but future studies might think of investigating each hypothesis specifically.