This study compares two approaches to the estimation of social welfare intervention costs: one "topdown" and the other "bottom-up" for a group of social welfare clients with severe problem behavior participating in a randomized trial. Intervention costs ranging over a two-year period were compared by intervention category (foster care placement, institutional placement, mentorship services, individual support services and structured support services), estimation method (price, micro costing, average cost) and treatment group (intervention, control). Analyses are based upon 2007 costs for 156 individuals receiving 404 interventions. Overall, both approaches were found to produce reliable estimates of intervention costs at the group level but not at the individual level. As choice of approach can greatly impact the estimate of mean difference, adjustment based on estimation approach should be incorporated into sensitivity analyses. Analysts must take care in assessing the purpose and perspective of the analysis when choosing a costing approach for use within economic evaluation.