The variation of oceanic basement roughness at mid-oceanic ridges is a complex trade-off between spreading rate that largely controls the thermal state of the lithosphere and its composition controlling the rheology and thus also the strength of the lithosphere. Here we estimate top basement roughness (i.e. the root-mean-square deviation of residual basement relief) over initial oceanic crust bordering the Iberia, Newfoundland, Bay of Biscay, Goban Spur, Flemish Cap, Australian and Antarctic rifted margins to provide new insights into the spreading processes at the nascent plate boundary. Although ultraslow seafloor spreading is suggested in those areas, the lack of undisputable oceanic magnetic anomalies prevents any well-constrained determination of the initial spreading rates. We compare these estimated roughness values with those determined over ultraslow-spreading crust formed at the Mid Atlantic Ridge, Southwest Indian Ridge, Arctic ridges, Mid-Cayman Spreading Center, Sheba ridge and South Pandora Ridge. The roughness values obtained at these ultraslow-spreading ridges range from 100 to >500 m and include 200-240 m roughness values which are typical of slow-spreading ridges. Roughness values larger than ∼300 m are characteristic of magma-poor sections of ultraslow-spreading ridges. The top basement roughness values determined within the inferred initial oceanic domain bordering the investigated magma-poor rifted margins are all higher than 200 m. Mean roughness values of the inferred initial oceanic domains alongside the conjugate Iberia and Newfoundland margins are greater than 300 m similarly to magma-poor sections of ultraslow-spreading ridges. The top basement in the initial oceanic crust alongside the conjugate Flemish Cap and Goban Spur margins shows roughness values and a tilted block morphology typical of slow-spreading ridges. We suggest that the roughness and the morphology of the top basement bordering the conjugate Australian and Antarctic margins indicate large tectonic extension and intermediate magma supply at either slow-or ultraslow-spreading rates. We show that estimating roughness values within transitional domains of magma-poor margins like exhumed mantle domains is less pertinent as polyphased tectonism and magmatism may have affected these domains leading to highly variable top basement roughness values that cannot be linked to a single process.