In order for high-fidelity radio wave propagation models to make accurate predictions a correct model of the refractivity is required. To this end, several methods have been developed to match refractivity profile structures between multiple range-dependent profiles. There is broad agreement that using refractivity to match features dismisses relevant information that should be used. The method presented here takes a different approach. When the parabolic equation methods are used to calculate propagation loss, the refractivity is only required at each range step. Hence any method that can correctly predict refractivity at the range step is appropriate. Previous methods match refractive modulus, M unit, features for a given vertical cut of the atmosphere, thus dismissing relevant information found in the temperature, pressure, and humidity profiles. Here we match features in three dimensions using the virtual potential temperature, a conserved quantity derived from temperature, pressure, and specific humidity. The variables used to compute refractivity are then interpolated in this matched coordinate system. This approach results in correct mapping of features over a region, resolving sampling issues and modeling refractivity with a smooth function in three dimensions. Therefore, as a parabolic equation steps in range, the refractivity is derived from the interpolated data. This paradigm eliminates the burden of refractivity interpolation/extrapolation from the propagation prediction software. Examples of matched atmospheres are presented, accompanied by propagation predictions based on the matched atmosphere. ¿ 2000 American Geophysical Union