The invention provides efficient methods of preparing a target nucleic acid in a form suitable for sequencing. The methods are particularly amenable for preparing high quality nucleic acids for massively parallel sequencing. The methods involve capturing a target nucleic acid from a sample and PCR amplification of the target nucleic acid. The target nucleic acid is captured by binding to a capture probe, which in turn binds to an immobilized probe. The immobilized probe is typically immobilized via a magnetic bead. The captured target nucleic acid is PCR amplified by thermocycling without prior dissociation of the target nucleic acid from the beads. The efficiency of the method lies in part in that both the capture and amplification steps are performed in a single vessel. The amplified nucleic acid can then be sequenced.