Accurate methods for the analysis of corneal topography are essential for the evaluation and management of corneal refractive errors. Photokeratoscopy has been used clinically to evaluate corneal shape anomalies, and methods have been developed to reconstruct the shape of the corneal surface from keratoscope photographs. However, the 2-dimensional information on the keratoscope image is insufficient for uniquely reconstructing a 3-dimensional corneal surface. Algorithms currently used make assumptions necessary for calculation, but lead to considerable error, especially at the periphery of radially aspheric corneas. This article proposes an improved algorithm that avoids some of the earlier assumptions for improved accuracy. The new algorithm was tested on both spherical and aspherical calibration surfaces. The result of this algorithm is identical to earlier methods for spherical surfaces, but the new method substantially increases accuracy in reproducing the aspherical surfaces at the cost of increased computation time.