The third build of the bovine genome comprises about 3 billion base pairs (bp; Ensembl, 2007). Locating functional DNA (fDNA) within this vast array of sequence is therefore a massive job. A method which can reduce the search space for this task may well have a wide application. It has been shown previously that the Neutral Indel Model (NIM; Lunter et al., 2006) can be used to locate a high proportion of human (Lunter et al., 2006) and bovine (Pollott, 2007) exonic DNA within a known 4% of the respective genomes. The NIM uses a mixture of methods from molecular evolution and comparative genomics to identify indel-purified segments (IPS) which should contain conserved fDNA. An IPS is the sequence located between two successive indels (insertions/deletions) and greater in length than a given threshold, as determined by input to the NIM method. Pollott (2007) found 64% of exonic DNA located in 233,715 IPS generated at a 0.1 false discovery rate (FDR) and these IPS overlapped with 87% of bovine genes. This increased to 81%, and 95%, respectively at a 0.5 FDR but the proportion of the genome searched increased from 3.3 to 10.7%. The objective of this work was to test some methods which may improve on the proportion of exonic DNA ‘found’ by the NIM without increasing the proportion of the genome searched by too much.