Child Maltreatment Fatalities (CMFs) have been alleged to exert great pressures on the delivery of Child Protective Services. Anecdotal accounts implicate CMFs as a cause of “vicious cycles” defined as sequences of scandal, outrage, and reform that destabilize and re-stabilize agency operations and personnel. In this paper we test the CMF cycle hypothesis using 22 years of data from the National Child Abuse and Neglect Data System (NCANDS). We employ Fourier time series and panel regression and find some evidence to support a cycle of CMF case accumulations that repeat at an 8 to 10 year interval. We show that previous levels of CMFs are much better predictors of subsequent CMFs than risk assessment factors used in much of the child welfare literature. We discuss the reasons why predictive analytics using traditional risk factors will continue to perform unsatisfactorily, and why dynamic models of CMFs will always perform better. Finally we consider how our research on the relative efficacy of traditional risk-as-hazard modelling vis a vis dynamic modelling can be used by academics and researchers to help practitioners operate more effective child welfare agencies.