Fluorescent nucleic acid hybridization probes traditionally have been generated by enzymatic incorporation of dye-labeled nucleotides, even though incorporation efficiency is low and variable from dye to dye. Alternatively, 5-(3-aminoallyl)-2'-deoxyuridine 5'-triphosphate (aa-dUTP) is enzymatically incorporated to generate amine-modified DNA, which is then chemically labeled with an amine-reactive fluorescent dye. We optimized this latter two-step approach for maximal hybridization signal brightness using DNA probes labeled to varying degrees with different fluorescent dyes. Reverse transcriptase and DNA polymerase 1 efficiently incorporated aa-dUTP into DNA, and adjusting the aa-dUTP:dTTP ratio controlled the degree of substitution. With cDNA probes hybridized to dot blots, probes having approximately eight dyes per 100 bases gave the best sensitivity, irrespective of the dye label. alpha-Satellite probes generated by nick translation and hybridized to human chromosome spreads also showed that probes having approximately eight dyes per 100 bases provided the brightest overall signals. These data demonstrate that this labeling method generates highly sensitive DNA probes that are difficult to obtain by conventional direct incorporation approaches. The technique is inherently consistent and versatile by virtue of the efficient incorporation of primary amines and the reliable chemical labeling reaction.