The study examines the criminal careers of male individuals born in 1940 (504), 1945 (515), 1950 (769), and 1955 (824) in Bayonne-Anglet-Biarritz. Attention is focused both on variations observed at the individual level (intracohort) and on variations resulting from sociological conditions specific to each cohort (intercohort). Information sources for the study are the documents and records from various agencies and institutions consulted. Six groups of variables are examined for the areas of social services, police, and justice. The evolution of data is interpreted according to a stochastic model which postulates that the probabilities of transition are constant. Thus, the probability that a subject will be in a given state at a given step is the same throughout the process. For each variable studied, matrices of state and of transition are established, and two types of statistical tests are carried out on the basis of those matrices. Results indicate that there is no influence of general sociological factors on crime and its development, that is, on the interactional dynamics between criminal behavior and social reaction, even where there are 20 years between cohorts. Criminal behavior and social reaction evolve according to different processes. Criminal behavior develops segmentally: each offense is only related to the immediately preceding offense, and no behavioral determinism is involved. In contrast, social reaction is a preestablished, not a sequential process, given direction especially by the necessities of institutional function. Thus, social stigmatization arises because the recidivism process is homogenous, while the social reaction system is not. The probability of offenders leaving the justice system becomes smaller at each progressive step of their careers. Transitions among various levels of the justice system are reduced as criminal involvement increases, so that virtually no contact exists between social services and repressive mechanisms. In fact, the social services have little contact with subjects of the type studied: only one in five has ever had treatment or assistance through social service institutions. Finally, patterns of criminal behavior characterized by certain tendencies for certain offenses may be observed, but careers are not consistent for particular crime types. It is concluded that cohort studies can be used successfully to trace criminal careers, using statistics in a meaningful fashion and introducing the clinical perspective. On a theoretical level, the study reaffirms the crucial importance of the clinical perspective in criminology, with its central focus on individual conduct and the personality which produces it. Notes are supplied.