The data used for the analysis of criminal careers were the FBI 'rap sheets' recording the arrests of 5,364 offenders who had 33,094 arrests. The group was chosen as a 100 percent sample of adult offenders who had experienced at least one arrest for an Index offense other than larceny (homicide, forcible rape, aggravated assault, robbery, burglary, or auto theft) during calendar 1973 in the District of Columbia. This crime-type selection criterion was specified to focus attention on individual offenders who had experienced at least one arrest for a reasonably serious offense. For the subjects in the sample, all prior adult arrests (after 18 years of age) were obtained. These arrests could have occurred anywhere in the United States and included any crime types (not only those six criterion crime types used for 1973 sample selection). Crime-switch matrices were used to determine any changes in offense seriousness in the course of the evolution of a criminal career. While it was anticipated that there would be a growth in the seriousness of offenses as an offender progressed into his criminal career, the study found that offense seriousness stabilized. Indeed, when the analysis controlled for record length (number of arrests), those with a longer record have a lesser average seriousness. A prospective analysis shows that the 'persisters,' those with at least three arrests, display a stable probability of having at least one more arrest, and this probability is stable for all record lengths. Thus, it appears that knowing the prior arrest history contributes very little in predicting future criminality. Tabular data and matrice equations are provided.