Long interspersed element 1 (LINE-1, L1) is a eukaryotic retrotransposon which copies itself through an RNA intermediate. The mutagenic insertion mechanism, termed target-primed reverse transcription (TPRT), requires coordinated activities of the encoded ORF2 protein (ORF2p) endonuclease (EN) and reverse transcriptase (RT) domains. TPRT is initiated when ORF2p EN nicks a target site in genomic DNA (gDNA), creating a 3'-OH that primes ORF2p RT for cDNA synthesis using the bound L1 RNA template. L1 insertions occur preferentially at 5'-TTTTT↓AA consensus motifs, a preference that could be driven by either site-specific EN cleavage or by sequence requirements in the subsequent RT priming step, in which the cut gDNA flap must base pair with the poly(A) RNA template. We find that in vitro, EN is promiscuous, cutting linear DNA oligos and plasmids at many non-consensus sites. We discovered a novel cleavage activity on a mismatched substrate that was nicked ∼40-fold faster than duplex DNA containing the consensus site and identify three features enabling rapid cutting. First, L1 EN cuts two nucleotides downstream of mismatches, favoring A-G mismatches or T•G/U•G wobble pairs. Second, both mismatch and consensus sequences are cut >2-fold faster when proximal to a DNA end. Third, end-proximal EN cutting depends on end composition: 5' overhangs cut fastest, followed by 3' overhangs, followed by blunt ends. These results suggest that EN cleavage is based primarily on DNA structure rather than sequence, that many attempted L1 insertions likely fail at the priming step after cleavage, and that DNA mismatches and possibly other alterations to DNA conformation promote EN cleavage, together broadening our understanding of the genomic impact of L1 expression.