IN a recent article Ferris (1986) offers a comprehensive, theoretically specified analysis of influences affecting the decision to contract out services in cities with populations equal to or greater than 25,000. In brief, he contends that since municipalities may not eagerly embrace contracting, preferring instead to maintain direct control over most services, the perceived gains from contracting must substantially exceed the potential costs. Three conditions likely to affect the benefit-cost calculation: when (1) cost savings likely; (2) fiscal pressures noticeable; and (3) political opposition is weak. To test his model Ferris identifies, justifies, and operationalizes 14 indicators of his three constructs: supply, fiscal, and political. The impact of the predictor variables on his dependent variable (the percentage of 43 publicly provided services that produced externally) is estimated through multiple regression analysis. Ferris (1986: 306) concludes that all three sets of factors [costs/supply, fiscal stress, and political considerations] instrumental in explaining the incidence of contracting out. Although Ferris's research must be considered one of the most ambitious attempts to date to explain the decision to contract out, his model can be extended and refined. For example, he does not examine any potential variation across major service providers, e.g., for-profit, nonprofit, and intergovernmental. Nor does he separate his contracting measure by functional area, e.g., public works, public safety, or parks and recreation. Previous research suggests that the propensity to contract varies across service vendors and by service area (see Ferris and Grady 1986; Valente and Manchester 1984). A third limitation of Ferris's contracting model is the decision to include as components of the dependent variable only those services delivered to the public [directly] (see p. 308, note 13). He excludes support services. As Gordon (1986: 226-27, emphasis in original) notes, traditional distinctions between line (direct) and staff (support) functions are increasingly coming to be seen as less distinct. Moreover, from a theoretical perspective staff-support services can significantly affect service efficiencies, a principal item of concern in Ferris's model.