The modification of glass surfaces with (3-mercaptopropyl)trimethoxysilane and the application of this to DNA chip technology are described. A range of factors influencing the silanization method, and hence the number of surface-bound, chemically active thiol groups, were investigated using a design of experiment approach based on analysis of variance. The number of thiol groups introduced on glass substrates were measured directly using a specific radiolabel, [14C]cysteamine hydrochloride. For liquid-phase silanization, the number of surface-bound thiol groups was found to be dependent on both postsilanization thermal curing and silanization time and relatively independent of silane concentration, reaction temperature, and sample pretreatment. Depending on the conditions used in liquid-phase silanization, (1.3-9.0) x 10(12) thiol groups/cm2 on the glass samples were bound. The reliability and repeatability of liquid- and vacuum-phase silanization were also investigated. Eighteen-base oligonucleotide probes were covalently attached to the modified surfaces via a 3'-amino modification on the DNA and subsequent reaction with the cross-linking reagent N-(gamma-maleimidobutyryloxy) succinimide ester (GMBS). The resulting probe levels were determined and found to be stoichiometric with that of the introduced thiol groups. These results demonstrate that silanization of glass surfaces under specific conditions, prior to probe attachment, is of great importance in the development of DNA chips that use the simple concept of the covalent attachment of presynthesized oligonucleotides to silicon oxide surfaces.