In May 2026, Rachel Entrekin won the Cocodona 250 outright, finishing first overall ahead of all male competitors in a 250-mile ultramarathon through the Arizona desert. Her victory was not anomalous: women increasingly appear in overall top-ten rankings at the longest distances, even as their participation rates decline sharply. This article investigates the gendered structure of trail running participation across 15,848 race editions in 95 countries (2010-2021), using the open-access TRAILGENDER database (Nakala DOI: 10.34847/nkl.cbbc2i86). We test five pre-registered hypotheses (OSF: osf.io/mwq92) addressing the moving glass ceiling, gendered self-selection, diffusion dynamics, structural exclusion thresholds, and continental convergence. Results reveal a striking 'ultra paradox': female participation decreases monotonically from 41.5% at 10K to 15.9% at extreme distances (ρ = -0.967), yet elite penetration increases inversely (ρ = +0.883). After introducing country fixed effects and nested multilevel modelling, two of five hypotheses require substantial revision: the apparent moving glass ceiling (H1) proves to be a compositional artefact, and continental variance (H5) is entirely absorbed by within-continent country heterogeneity. A structural exclusion threshold at approximately 25 finishers (H4) remains robustly supported, consistent with Schelling's critical mass theory. We propose a three-level self-selection modelcountry engagement, intensity filtering, and elite penetration-that reframes the gender gap as a cascading filtration process rather than a uniform barrier. These findings challenge both deficit-based narratives of female under-representation and celebratory accounts of ultra-distance convergence, offering instead a structural account with direct implications for sport governance.

In May 2026, Rachel Entrekin won the Cocodona 250 outright, finishing first overall ahead of all male competitors in a 250-mile ultramarathon through the Arizona desert. Her victory was not anomalous: women increasingly appear in overall top-ten rankings at the longest distances, even as their participation rates decline sharply. This article investigates the gendered structure of trail running participation across 15,848 race editions in 95 countries (2010-2021), using the open-access TRAILGENDER database (Nakala DOI: 10.34847/nkl.cbbc2i86). We test five pre-registered hypotheses (OSF: osf.io/mwq92) addressing the moving glass ceiling, gendered self-selection, diffusion dynamics, structural exclusion thresholds, and continental convergence. Results reveal a striking 'ultra paradox': female participation decreases monotonically from 41.5% at 10K to 15.9% at extreme distances (ρ = -0.967), yet elite penetration increases inversely (ρ = +0.883). After introducing country fixed effects and nested multilevel modelling, two of five hypotheses require substantial revision: the apparent moving glass ceiling (H1) proves to be a compositional artefact, and continental variance (H5) is entirely absorbed by within-continent country heterogeneity. A structural exclusion threshold at approximately 25 finishers (H4) remains robustly supported, consistent with Schelling's critical mass theory. We propose a three-level self-selection modelcountry engagement, intensity filtering, and elite penetration-that reframes the gender gap as a cascading filtration process rather than a uniform barrier. These findings challenge both deficit-based narratives of female under-representation and celebratory accounts of ultra-distance convergence, offering instead a structural account with direct implications for sport governance.