W e have shown previously that periodate oxidation of collagen carbohydrate does not affect its ability to aggregate platelets. We now describe an additional characterization of periodate-modified collagen which demonstrates that collagen devoid of intact carbohydrate is fully capable of fibril formation, and we confirm its capacity to initiate platelet aggregation. Furthermore, we demonstrate that the platelet aggregating abilities of Types I, II, and III fibrillar collagen are quite similar despite differences in carbohydrate content and amino acid sequence. We also demonstrate that monomeric, pepsin-solubilized Type I human collagen is ineffective inhibiting aggregation by performed fibrils derived from the same molecule, thus establishing that the affinity of platelets for collagen depends upon prior polymerization of collagen. We interpret these and other findings to demonstrate that the hydroxylysyl glycoside regions of collagen are not highly specific sites involved in platelet-collagen interactions leading to "physiological" aggregation, and that the possibility must be considered that multiple interactions involving collagen sites of comparatively low structural specificity may be the initiating events in release of platelet ADP and the ensuing aggregation.