This paper investigates the technology cost structure in US physical therapy care. We exploit formal economic theories and a rich national data of providers to tease out implications for operational cost efficiencies. The 2008–2009 dataset comprising over 19 000 bi-weekly, site-specific physical therapy center observations across 28 US states and Occupational Employment Statistics data (Bureau of Labor Statistics) includes measures of output, three labor types (clinical, support, and administrative), and facilities (capital). We discuss findings from the iterative seemingly unrelated regression estimation system model. The generalized translog cost estimates indicate a well-behaved underlying technology structure. We also find the following: (i) factor demands are downwardly sloped; (ii) pair-wise factor relationships largely reflect substitutions; (iii) factor demand for physical therapists is more inelastic compared with that for administrative staff; and (iv) diminishing scale economies exist at the 25%, 50%, and 75% output (patient visits) levels. Our findings advance the timely economic understanding of operations in an increasingly important segment of the medical care sector that has, up-to-now (because of data paucity), been missing from healthcare efficiency analysis. Our work further provides baseline estimates for comparing operational efficiencies in physical therapy care after implementations of the 2010 US healthcare reforms. Copyright © 2012 John Wiley & Sons, Ltd.