The idea of a simple linear boundary between continental and oceanic crust at extended continental margins is widely recognized to be an oversimplification. Despite this, such boundaries continue to be mapped because of their perceived utility in palinspastic and plate kinematic reconstructions. To examine whether this perception is justified, we review the data and models on which basis continent ocean boundaries are interpreted, and map a set of such interpretations worldwide from more than 150 publications. The maps show that the location of the continent ocean boundary is rarely consistently estimated within the ~ 10–100 km observational uncertainty that might be expected of the geophysical data used for doing so, that this is the case regardless of whether the transition zone behind the boundary is classified as magma rich or magma poor, and that the geographical separation of estimates exceeds the width of single-study continent ocean transition zones. The average of global maximum separations across sets of three or more estimates is large (167 km) and mostly a consequence of interpretations published over the last decade. We interpret this to indicate an extra component of uncertainty that is related to authors' understanding of the range of features that are interpretable at extended continental margins. We go on to discuss the implications of this uncertainty for palinspastic and plate kinematic modelling using examples from the literature and from the South Atlantic Ocean. We conclude that a precise continent ocean boundary concept with locational uncertainty defined from the ensembles is of limited value for palinspastic reconstructions because the reconstruction process tends to bunch the ensemble within a region that is (i) of similar width to the observational uncertainties associated with continent ocean boundary estimates, (ii) narrower than the regions of uncertainty about rotated features implied by the propagation of uncertainties from plate rotation parameters, and (iii) coincident, within all the above uncertainties, with the more-easily mapped continental shelf gravity anomaly. Secondly, we conclude that estimated continent ocean boundaries are of limited use in developing or testing plate kinematic reconstructions because (i) reconstructions built using them as markers do not, within uncertainty limits defined from the ensembles, differ greatly from those using more-easily determined bathymetric or gravity anomaly contours, and (ii) because it is impossible to segment and date them with useful precision to use as markers of the edges of rigid oceanic lithosphere outside of the constraints of a pre-existing plate kinematic model.