The polymerase chain reaction (see Chapter 1) allows the rapid isolation of specific DNA targets that may be used as sequencing templates either directly or after cloning into M13. The latter procedure allows single-strand sequencing, but is otherwise undesirable not only because it is slow, but also because a significant proportion of the amplified DNA molecules contain replication errors. These are expected to occur at a frequency of 1 in 10,000 bases incorporated (1), and will also be amplified during subsequent cycles. This means that at least three clones from independent amplification experiments must be sequenced in order to identify these replication errors and determine the final consensus sequence. Direct sequencing of the PCR product bypasses this problem since it produces an “average sequence” of all the copies of the target, and any miscopied molecule is bound to represent only a very small proportion of the total (unless one starts PCR with very few molecules of the target DNA).