It was found that these criminal careers could be modeled with parameters reflecting constant individual rates of offending and constant probability of career termination; population heterogeneity could be adequately represented by two distinct groups -- designated as frequents and occasionals. In the London cohort, the frequents were estimated to have an annual conviction rate of 1.14 (constant with age) and a probability of career termination of .10 following each conviction. The occasionals had an annual conviction rate of .41 and a termination probability of .33 following each conviction. The frequents were estimated to comprise 43 percent of the population; the occasionals, 57 percent. While this parsimonious model structure was adequate for this sample, it requires testing with other offender populations. 7 tables and calculations are appended. 13 references. (Author abstract modified)