The study found that all four offenses were associated with a heightened likelihood that the offender would commit a subsequent homicide compared to the general population. Those who engaged in kidnapping had a higher likelihood of committing a subsequent homicide during a 20-year followup than those convicted of the other crimes (1.0 percent). Those who committed arson, blackmail, and threats to kill had a similar risk of committing a subsequent homicide (0.8 percent) during the 20-year followup. The sequencing of the offenses was also relevant to the commission of a subsequent homicide. Those convicted of more than one type of serious offense were at higher risk of a subsequent homicide conviction. There was also evidence of specialization, particularly for arsonists, among serious offenders who reoffended. The study obtained data from the Offenders Index, an administrative dataset that consists of records of convictions for all standard list offenses for all offenders in England and Wales since 1963. The datasets were constructed in two stages, the first stage identified all offenders convicted of arson, blackmail, kidnapping, or threats to kill between 1979 and 2001. Four separate datasets, each for one of the offenses were developed. The second stage combined the four datasets into a single dataset. The construction of this dataset allowed the examination of temporal sequences of serious-crime convictions and the subsequent risk of first-time homicide following these sequences. 5 tables, 2 figures and 12 references