Surface-based analysis is a popular approach to fMRI analysis as, compared with volumetric analysis, it allows easier registration of functional regions between individuals. This work sought to move towards surface-based perfusion analysis of arterial spin labelling data. It presents a comparison of the results of a surface-based analysis pipeline with results from a conventional volumetric perfusion analysis pipeline projected onto the cortical surface. Bayesian inference of cortical surface perfusion is demonstrated here and is shown to produce substantially lower perfusion estimates than conventional voxelwise analysis. Future work will investigate to origins of these perfusion differences.