Conventional theories of crime rate trends explain that crime rates progress linearly, which ensures the predictability of the future from the past. An emerging class of historical contingency theories has questioned the conventional assumptions about crime rates, claiming instead that the process underlying crime rates is nonlinear. In order to test whether the historical contingency theories offer an improvement over the conventional theories of crime rate trends, the authors analyzed United States crime rates over the last two-thirds of the 20th century, testing separately for time irreversibility, gradual parameter variation, and abrupt structural breaks. The results indicate none of the features one would expect if the historical contingency theories are accurate. The data on homicides, robberies, and assaults did not deviate from the constant Gaussian linear process. Thus, the conventional linear models appear adequate to explain most crime rate trends. The findings, however, highlight an important gap in the formulation of conventional theories in that they point to the nonstationary nature of crime rate levels. Future research should focus on locating the source of the nonstationarity of crime rates. Figure, tables, notes, references