Traditional methods of Curie depth estimation involving spectral analysis of low-level aeromagnetic data require that depth to the bottom of a magnetic crustal layer be larger relative to depth to top, so that spectral separation can be achieved. An alternate method for high-altitude data, where depth to magnetic layer top and bottom are virtually indistinguishable, is described. Apparent magnetization variation in a thin equivalent layer is obtained by inversion of the anomaly data. The information in such a model is essentially the vertical integral of magnetization in the real layer to some resolution limit and within some ambiguity in level. This can be transformed to a model of thickness variation in a layer of constant magnetization, given some minimal constraints. The thickness variations may reflect Curie isotherm undulation in regions of high crustal heat flow; Curie depth models then provide a constraint on crustal thermal models. The method was first applied to a magnetization model for the western United States derived from Magsat data. A Curie surface model was obtained and used to constrain simple thermal models along two crustal sections. Associatd surface heat flow is in agreement with observations. The method was then applied to upward continued aircraft data from the Oregon Cascades. Results are in reasonably good agreement with previous results based on spectral analysis and with measured heat flow.