There are several methods presently employed to estimate developmental trajectories, such as hierarchiecal linear modeling, growth curve analysis, and semiparametric mixture models. This paper focuses on the semiparametric mixture model (or latent class analysis) which is a group-based approach employed by a growing body of research to discover underlying developmental trajectories of crime. Due to this method’s growing popularity, it is believed that researchers need to assess the robustness of the trajectory attributes under varying conditions. Three methodological considerations were used to examine how trajectories of offending were affected by varying conditions: (1) length of follow-up; (2) the inclusion of incarceration information to account for exposure time on the street; and (3) the inclusion of mortality information to safeguard against treating those who are dead as desisters from crime. The analysis included 1,000 males from Boston, MA, and who were born between 1925 and 1932. However, this study focused on 500 juvenile delinquent males from the total sample of 1,000. Overall, the analysis investigates how the varying conditions of data outlined can affect trajectory patterns produced by the semiparametric mixed Poisson model. The conclusion drawn was that these data conditions did influence the results of the analysis, sometimes in significant ways. More precisely, longer-term data on offending and the inclusion of incarceration and mortality information appeared to be key pieces of information, especially when analyzing high-rate offending patterns. Researchers are encouraged to conduct future sensitivity analyses of the semiparametric mixture model method, as well as others. Figures and references