Spatial syntax analysis has become an influential method of analysing street networks as spaces of pedestrian movement. While significant correlations have been found between pedestrian flows and axial topological models, these models are inconsistent in the way urban morphologies are represented and measured. Meanwhile, in the fields of health, transport and urban design research, correlations have been found between walking and a range of urban morphological attributes that topological models ignore. This paper shows that while space syntax analysis has become increasingly sophisticated over the past decades, substantial limitations persist. Focusing on the limits of the theory in its own substantive field, it is shown that the abstraction of the street to its axial line poses three fundamental problems. First, it eliminates the street section and thus does not recognise that the social logic of space is also transversal across the street. Second, it ignores permeability as a key morphological attribute linked to walkability at neighbourhood scale. Third, it transposes smooth urban conditions into striated measurable models that iron-out ambiguities, eliminating conditions of liminality, porosity and complexity. Yet all these dimensions have been recognised as key attributes of urban intensity at street level. In conclusion it is argued that while axial integration may be useful in studying larger urban networks to capture particular morphogenetic tendencies, it can be misleading as a measure of walkable access at neighbourhood scale.