Apparent diffusion coefficient (ADC)-fMRI offers a promising functional contrast, capable of map-ping neuronal activity directly in both grey and white matter. However, previous studies have shown that diffusion-weighted fMRI (dfMRI), from which ADC-fMRI derives, is influenced by BOLD effects, leading to a concern that the dfMRI contrast is still rooted in neurovascular rather than neuromor-phological coupling. Mitigation strategies have been proposed to remove vascular contributions while retaining neuromorphological coupling, by: i) analysing ADC timecourses calculated from two inter-leaved diffusion-weightings, known as ADC-fMRI; ii) using b-values of at least 200 s mm-2; and iii) using a sequence compensated for cross-terms with fluctuating background field gradients associated with blood oxygenation. Respiration-induced haemodynamic fluctuations, which are dissociated from neural activity, are an excellent test-bed for the robustness of ADC-fMRI to vascular contributions. In this study, we investigate the association between end-tidal CO2 and ADC-fMRI, in comparison with dfMRI and BOLD, in both breath-hold and resting-state paradigms in the human brain. We confirm a strong dependence of the BOLD signal on respiration, and find a pattern of delayed haemodynamic response to respiration in regions comprising the default mode network. While dfMRI mitigates much of the vascular contribution, it retains some association with respiration, as expected. Conversely, ADC-fMRI is mostly unaffected by vascular contribution, exhibiting minimal correlation between ex-pired CO2 and ADC timeseries, as well as low inter-and intra-subject reproducibility in correlation maps. These findings validate ADC-fMRI as a predominantly non-vascular contrast sensitive to mi-crostructural dynamics, enabling whole-brain functional imaging unconstrained by vascular confounds. Significance statement Apparent diffusion coefficient (ADC)-fMRI offers a unique neuromorpho-logical contrast, enabling functional imaging of both grey and white matter with greater spatial and tem-poral specificity than BOLD, as demonstrated in task and resting-state studies. Recent methodological advances have isolated ADC-fMRI from confounding haemodynamic influences, increasing specificity to microstructural neuronal dynamics. Here, we provide the first evaluation of these strategies in breath-hold and resting-state conditions, monitoring respiration as a proxy for haemodynamic fluctuations dissoci-ated from neural activity. Our results show minimal association between ADC-fMRI and respiration, confirming that ADC-fMRI is largely immune to haemodynamic fluctuations. It provides crucial vali-dation that ADC-fMRI reflects a non-vascular functional contrast distinct from and complementary to BOLD, supporting its utility in mapping activity across the human brain.