Studies of seismic variability play a more important role in the understanding of processes that form oceanic crust than individual velocity-depth solutions found in isolation. For such variability studies, solutions must be contrasted with one another, this is normally not straightforward owing to the lack of rigorous uncertainty estimates available for solutions. Here I present a method for the direct comparison of such solutions by hpothesis testing. A null hypothesis is proposed that a single velocity solution is consistent with two sets of data. The hypothesis is tested by seeking such a model using waveform inversion and comparing its misfit against the misfit from fully independent solutions. The degree of similarity of the velocity structures can be further quantified by solving for the pair of solutions that are as similar as possible while consistent with each set of data; the F test is used to evaluate the the significance of each additional effective model parameter as it lends independence to the two solutions. The method processes advantages of a hypotheical Monte Carlo search of all modes because the complete forward problem is carried out at each step, yet it is tractable because the search direction is well defined. The method is applied to data sets collected near the Mid-Atlantic Ridge on 0 and 7 m.y. crust in order to quantify the differences between the solutions. The comparison finds quantifiable differences in both layers 2 and 3 between these two regions, suggesting that significant evolution of the entire crustal section has taken place in the first 7 m.y. ¿American Geophysical Union 1992