To track the processing of damaged DNA double-strand break (DSB) ends in vivo, a method was devised for quantitative measurement of 3'-phosphoglycolate (PG) termini on DSBs induced by the non-protein chromophore of neocarzinostatin (NCS-C) in the human Alu repeat. Following exposure of cells to NCS-C, DNA was isolated, and labile lesions were chemically stabilized. All 3'-phosphate and 3'-hydroxyl ends were enzymatically capped with dideoxy termini, whereas 3'-PG ends were rendered ligatable, linked to an anchor, and quantified by real-time Taqman polymerase chain reaction. Using this assay and variations thereof, 3'-PG and 3'-phosphate termini on 1-base 3' overhangs of NCS-C-induced DSBs were readily detected in DNA from the treated lymphoblastoid cells, and both were largely eliminated from cellular DNA within 1 h. However, the 3'-PG termini were processed more slowly than 3'-phosphate termini, and were more persistent in tyrosyl-DNA phosphodiesterase 1-mutant SCAN1 than in normal cells, suggesting a significant role for tyrosyl-DNA phosphodiesterase 1 in removing 3'-PG blocking groups for DSB repair. DSBs with 3'-hydroxyl termini, which are not directly induced by NCS-C, were formed rapidly in cells, and largely eliminated by further processing within 1 h, both in Alu repeats and in heterochromatic α-satellite DNA. Moreover, absence of DNA-PK in M059J cells appeared to accelerate resolution of 3'-PG ends.