The development of a red cell substitute by chemical modification of hemoglobin has been approached as a systematic, iterative process. Acyl phosphate methyl esters were designed as anionic electrophiles to permit selective acylation of amino groups in the cationic site of hemoglobin which binds polyanions. Kinetic studies with systematically substituted acyl phosphates and amines show that the reaction is controlled by a reversible addition step followed by an irreversible elimination step. Acyl phosphate methyl esters which are derivatives of rigid dicarboxylic acids introduce cross-links in human hemoglobin between amino groups in the beta subunits (epsilon-NH2-Lys-82, alpha-NH2-Val-1) and permit correlation of oxygen binding properties with cross-link structure. The data suggest that the cross-link maintains cooperativity while reducing overall oxygen affinity by lowering the affinity of the R form for oxygen rather than by perturbing the R,T equilibrium of native hemoglobin. Materials produced from deoxyhemoglobin with a cross-link between positions 1 and 82 of the two beta units have appropriate oxygen affinity for red cell substitutes. The use of a trifunctional cross-linker, trimesyl tris(methyl phosphate) selectively produces hemoglobin with the desired 1-82 connection in good yield. The reagent is readily prepared and the properties of this chemically modified hemoglobin are suitable for trial as a red cell substitute, closely resembling those of optimized materials produced by recombinant technology. Further work is producing new chemicals and providing structural information.