Sublimation modifies the stable isotopic composition of snow. The physical controls governing this imprint, particularly the interplay between isotopic fractionation, snow structure, and airflow, remain poorly constrained. Current surface-exchange models commonly treat sublimation fractionation as a function of atmospheric forcing and boundary-layer exchange, with limited explicit representation of snow structural state. Here, we present controlled wind-tunnel experiments examining sublimation-driven isotopic changes (δ¹⁸O, δ²H, and d-excess) in three contrasting snow types (laboratory snow, fresh natural snow, and old natural snow) under varied airflow regimes, ambient humidity, and basal-to-air temperature contrasts. We show that airflow is the dominant experimental control on sublimation mass loss, but bulk isotopic change does not scale directly with mass loss, revealing a partial decoupling between sublimation magnitude and isotopic modification. High airflow produced the smallest isotopic change per unit mass loss, consistent with strong turbulent exchange reducing the isotopic imprint of sublimation. However, absolute isotopic change varied systematically with snow structural state, scaling strongly with bulk porosity in natural snow. This structural dependence weakened after mass normalization for δ¹⁸O and δ²H, but persisted for d-excess, suggesting that properties associated with porosity influence effective kinetic fractionation in addition to cumulative isotopic change. Laboratory snow deviated from this behavior: despite overlapping the porosity range of natural snow, it showed no porosity-isotopic change relationship. This suggests that bulk porosity alone is insufficient to capture the structural controls on isotopic response to sublimation. Within a relative snow-type comparison, the Craig–Gordon model systematically overpredicted δ¹⁸O and δ²H enrichment and underpredicted the range of d-excess change. The systematic nature of these discrepancies suggests that parameter uncertainty alone is unlikely to explain the mismatch and points to missing representation of internal vapor transport through the snow pore network. Our study shows that sublimation rate alone is insufficient to predict isotopic modification; snow structural state must be considered alongside atmospheric forcing in models of sublimation-driven isotopic evolution. Snow structural state can therefore impose a systematic overprint on isotopic signals in snow archives and meltwater that structure-independent models are likely to misrepresent.