The membranolytic activity of silica particles toward red blood cells (RBCs) has been known for a long time and is sometimes associated with silica pathogenicity. However, the molecular mechanism and the reasons why hemolysis differs according to the silica form are still obscure. A panel of 15 crystalline (pure and commercial) and amorphous (pyrogenic, precipitated from aqueous solutions, vitreous) silica samples differing in size, origin, morphology, and surface chemical composition were selected and specifically prepared. Silica particles were grouped into six groups to compare their potential in disrupting RBC membranes so that one single property differed in each group, while other features were constant. Free radical production and crystallinity were not strict determinants of hemolytic activity. Particle curvature and morphology modulated the hemolytic effect, but silanols and siloxane bridges at the surface were the main actors. Hemolysis was unrelated to the overall concentration of silanols as fully rehydrated surfaces (such as those obtained from aqueous solution) were inert, and one pyrogenic silica also lost its membranolytic potential upon progressive dehydration. Overall results are consistent with a model whereby hemolysis is determined by a defined surface distribution of dissociated/undissociated silanols and siloxane groups strongly interacting with specific epitopes on the RBC membrane.