Probabilistic tsunami hazard analysis (PTHA) due to submarine landslides is much less developed than PTHA for earthquake sources. This is partly because of less constrained data on source probability, and partly due to lack of knowledge related to the tsunami generation process due to landslide dynamics. This study provides a basis for estimating the uncertainty related to landslide dynamics for PTHA from submarine landslides based on a new landslide database in the Gulf of Cadiz. The establishment of this new database is described herein. We use submarine landslide run-out statistics from this database to calibrate landslide parameters and related uncertainties using the cohesive landslide model BingClaw. In turn, coupling the landslide motion to tsunami genesis is used to characterise the inferred tsunami uncertainties. Important parameters that can explain the large tsunami uncertainties are the initial water depth of the landslide and the slope, the landslide volume, and the initial yield strength of the landslide material. Kinematic properties such as the initial landslide acceleration or the Froude number are found to strongly correlate with tsunami genesis. In this study, we show how the fitting process of numerical models landslide run-out can be casted into uncertainty in maximum sea surface elevations heights. This can in turn be an to a future PTHA for spanning uncertainty ranges due to the landslide dynamics on tsunami-genesis, constrained by landslide run-out data.