The effects of urea on apomyoglobin solubility have been investigated. Apomyoglobin precipitation was found to be a thermodynamically reversible process independent of the pathway of aggregation. A liquid-solid phase diagram was constructed for the precipitation of apomyoglobin as a function of urea and protein concentration. Apomyoglobin solubility decreases by an order of magnitude between 0 and 1.5 M urea, reaching a minimum near 2.4 M urea and increasing at higher urea concentrations (the denaturation midpoint is at approximately 2.6 M urea). This decrease in protein solubility is opposite to that expected based on amino acid solubilities, since both polar and nonpolar molecules become more soluble with increasing urea concentration. Solubility minima for proteins have been rationalized in terms of folding intermediates. However, our structural studies show no evidence for folding intermediates in apomyoglobin under the experimental conditions, apart from small predenaturation changes. Our data are consistent with an alternative hypothesis, namely, that the primary aggregating species are denatured protein molecules, rather than intermediate states. Consistent with recent thermodynamic and statistical mechanical models, the solubility minimum may be described as the result of two competing effects of urea: (1) urea denatures the protein, and (2) urea makes the solvent more favorable for the native and any denatured state. At low urea concentration, solubility decreases with increasing urea concentration due to the domination of the solubility behavior by the increase in the population of aggregation-competent (denatured) protein molecules. However, at high urea concentration, the increasingly favorable nature of the solvent dominates, resulting in increasing solubility with urea concentration. The phase diagram provides guidance for the best experimental conditions (pathway) to use to avoid aggregation during the refolding of denaturant-unfolded protein.