The origin of the Caribbean plate - Pacific or intra-American - is still under debate. We produced a magnetic map from our compilation of marine magnetic data in the Caribbean plate and the Gulf of Mexico (GoM) in order to perform a magnetic interpretation which could contribute to the debate. In the GoM, we relate a set of fan-like long-wavelength magnetic anomalies with seafloor spreading. We interpret these anomalies by comparing them to filtered polarity time scales and identify the fossil ridge axis. We then carry out plate tectonic reconstruction to establish the initial tectonic framework and evolution of the GoM. The GoM opened asymmetrically in two stages, starting during the Kimmeridgian and ceasing in the Berriasian. We interpret the strong magnetic anomalies and neighboring smoothed magnetic anomalies extending from South to North in the Colombian Basin as Chrons 33 and 33r and the younger part of the CQZ. Comparing the width of magnetic anomalies 33 and 33r in the Colombia Basin, Atlantic and Pacific Oceans shows a similarity between those in the Colombian Basin and the Pacific Ocean. Forward modelling in the Colombian Basin allowed to propose a paleo-latitude of ∼10°S and a spreading rate of ∼3.6 cm/yr. The crust in the Venezuela Basin shows North-South fracture zones and stronger magnetic variations that would be older, belonging to the middle part of the CQZ. The Caribbean plate more likely formed on the Pacific-Farallón spreading center.