In situ bioremediation is being considered to optimize an existing pump-and-treat remedy for treatment of explosives-contaminated groundwater at the Umatilla Chemical Depot. Push-pull tests were conducted using a phased approach to measure in situ hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) and 2,4,6-trinitrotoluene (TNT) degradation rates associated with various carbon substrates. Phase I included short-duration transport tests conducted in each well to determine dilution rates and retardation factors for RDX and TNT. Phase II included aquifer “feedings” conducted by injecting 150 gallons of treated site groundwater amended with ethanol, corn syrup, lactose or emulsified oil (concentrations 10, 25 and 27 mM, respectively; 12% by volume for emulsified oil). Wells received up to 6 substrate “feedings” over the course of 3 months followed by monitoring dissolved oxygen, nitrate, Fe(II), and sulfate to gauge in situ redox conditions as indicators of anaerobic microbial activity. Phase III included push-pull tests conducted by injecting 150 gallons of site groundwater amended with approximately 1000 µg/L RDX, 350 µg/L TNT, carbon substrate and a conservative tracer, followed by sampling over 8 d. Corn syrup resulted in the best RDX removal (82% on average) and the largest RDX degradation rate coefficient (1.4 ± 1.1 d−1). Emulsified oil resulted in the best TNT removal (99%) and largest TNT degradation rate coefficient (5.7 × 10−2 d−1). These results will be used to simulate full-scale in situ bioremediation scenarios at Umatilla and will support a go/no-go decision to initiate full-scale bioremediation remedy optimization.