The determination of an initial model for the crystal structure of des-pentapeptide in-sulin, using only the phase information obtained from the scattering of cadmium ions inthe native crystal, has been described. This paper describes the refinement of this modelagainst both the experimental X-ray observations and the a priori stereochemical observationsusing the method of restrained parameter least squares. The corrected and extended modelincluded not only all the non-hydrogen atoms of the protein and one cadmium ion, but alsosome 49 water molecules. Fourier syntheses using amplitudes of 2F_o-F_c, phase angles prob-abilistically combined from the anomalous scattering and model phase information, witheach term weighted according to its reliability gave clear results without detectable bias to-wards the current model. The final model has acceptable stereochemistry (root mean squaredeviation from ideal bond lengths 0.027 A), and predicts the observed X-ray scattering sat-isfactorily (R = 0.182 for all dat The determination of an initial model for the crystal structure of des-pentapeptide in-sulin, using only the phase information obtained from the scattering of cadmium ions inthe native crystal, has been described. This paper describes the refinement of this modelagainst both the experimental X -ray observations and the a priori stereochemical observations using the method of restrained parameter least squares. The corrected and extended modelincluded not only all the non-hydrogen atoms of the protein and one cadmium ion, but alsosome 49 water molecules. Fourier syntheses using amplitudes of 2F_o -F_c, phase angles prob-abilistically combined from the anomalous scattering and model phase information, witheach term weighted according to its reliability gave clear results without detectable bias to-wards the current model. The final model has acceptable stereochemistry (root mean square deviation from ideal bond lengths 0.027 A), and predicts the observed X-ray scattering sat-isfactorily (R = 0.182 for all dat