tRNAscan-SE uses an optimized version of tRNAscan 1.3 [Fichant & Burks, 1991] which we refer to as tRNAscan 1.4. The core algorithm is identical to tRNAscan 1.3. tRNAscan versions 1.3 and 1.4 have identical tRNA detection rates except in the case of ambiguous nucleotides occurring within the input sequence. There are implementation errors in tRNAscan 1.3's handling of ambiguous nucleotide codes. tRNAscan 1.4 conservatively calls ambiguous nucleotides as always forming base pairings in stems, and matching the highest scoring choice in consensus promoter matrices. This results in a high false positive rate for sequences containing a large number of ambiguous nucleotides. For our purposes, this is acceptable because the second stage covariance model analysis eliminates false positives. Several command line options were added to tRNAscan 1.4 for convenience in integration with tRNAscan-SE. Additional code changes were made to increase the robustness and speed of the program. These modifications result in roughly a 650-fold increase in search speed and no upper limit on input sequence size.