This dissertation builds an empirical justification for the application of systems theory in child welfare research, policy, and practice. This dissertation considers the theoretical question of whether systems structures shape child welfare populations. Synthesizing population ecology theory and social dynamics theory, this study predicts that patterns of entries into and exits out of the child welfare system will manifest themselves differently if entries and exits are constrained by system structures as opposed to if those cases are reviewed individually and independently.,Applying that theory, this dissertation then considers the empirical question of whether observing, in this case using Empirical Dynamical Modelling, the dynamics of child welfare populations over time allows those structures to be inferred. This study focuses on quantitative analysis of child welfare administrative data collected in six Washington State counties between the years 2000 and 2014 within two child welfare subsystems: out-of-home care and dependency court. In order to better contextualize the quantitative analysis in this study, a limited qualitative inquiry focused on bounding the analysis in a meaningful, practice-relevant scope was conducted.,Analysis found some support for coupling of entry and exit behavior both within and between child maltreatment court and out-of-home care subsystems. Relationships were stronger within subsystems than between subsystems. Analysis found nonlinearity across subsystems. Prior to the quantitative analysis, conducted qualitative interviews with thirteen practitioners in two Washington counties identified conditions which could produce feedback, capacity, and rate governing behavior in child welfare systems. Importantly, practitioner reports focused the quantitative analysis on week- and county-level aggregations for quantitative analysis.,This dissertation contributes to social work by conceptualizing and empirically justifying a population-based frame to develop systems thinking in child welfare. Further, this dissertation identifies nonlinearities and coupled relationships in child welfare time series which require methodological reconsideration in child welfare analysis.