Knowledge of the mechanisms and controls of ice stream flow is fundamental to understanding ice discharge from large ice sheets. Here, we utilise an extensive (9950 km2) marine geophysical dataset (comprising multibeam swath bathymetry, sub-bottom and seismic profiles) to describe the geomorphological signature of a 265 km-long, 65 km-wide palaeo-ice stream in the western Amundsen Sea Embayment, West Antarctica. We show that substrate was a dominant control on the ice streams relict bedform morphology. Data illustrate a large convergent streaming system, which drained through a cross-shelf trough at the Last Glacial Maximum (LGM), and probably during past glaciations. The sea-floor geomorphic imprint of the ice stream comprises more than 4000 elements, which have been digitised and mapped in a GIS. These include mega-scale glacial lineations (MSGLs), drumlins, meltwater channels, and grounding zone wedges. Bed roughness analysis and seismic profiles show that the former ice substrate was characterised by regions of rugged acoustic basement on the inner shelf (consisting of bedrock or indurated sedimentary rocks) and relatively smooth, truncated sedimentary strata, on the mid-to-outer shelf. A landsystem interpretation of the superimposed bed signature reveals a strong correlation to the substrate type, which can be sub-divided into five landsystem components: (1) a meltwater assemblage, (2) a composite ice sheet assemblage, (3) a sub-ice stream footprint group, (4) grounding line retreat morphology, and (5) a pro-marginal deglacial group. The MSGLs record a simple snapshot of the sole of the ice stream flowing over sediments, towards the end of the last glaciation, and although they are masked by iceberg ploughmarks on the outer shelf, we infer an extensive ice sheet in this sector at the LGM. By contrast, detailed landform mapping of the inner shelf convergence zone demonstrates overprinting, preservation and a time-transgressive history for the inner shelf landform groups, implying that bedforms, when viewed together, cannot be considered part of a single down-flow continuum, as suggested previously. Variability among inner shelf bedform types, their geometry and elongation also reveals information about local, complex basal-ice conditions and dynamics, not apparent at the scale of regional morphological studies, and which heavily influenced the bedform imprint. Thus, while we believe the analysis of a complete morphological dataset is a robust method for interpreting broad and localised palaeo-ice flow characteristics from bed signatures, we suggest that direct inferences about the dynamics and velocity of former ice streams, from limited bedform information, should be treated with caution.