Flow-routing studies were made to evaluate the response of the Lehigh and Delaware Rivers to low-flow augmentative releases from two reservoirs --Francis E. Walter Reservoir and Beltzville Lake--in the Lehigh River basin. Digital routing models that use diffusion-analogy methods to convolute flows with system-response functions were developed to simulate daily flows at selected sites. Model errors, for five sites and for periods of 1 year or more, were mostly between 3 and 12 percent in terms of absolute errors in daily flows and were mostly within 4 percent for flow volumes. The developed models were satisfactory for predicting hydrographic response at eight sites in the reach from White Haven, Pennsylvania to Trenton, New Jersey. However, abrupt changes in the flow rate of the Lehigh River at the Bethlehem and the Glendon gaging stations could not be adequately replicated with the model. The model tends to underestimate peaks by as much as 30 percent and to overestimate some low flows of short duration by as much as 20 percent. This occurs primarily because inflows from ungaged areas could not be reliably modeled throughout their ranges by use of flow records for gaged streams. The model will underestimate long-duration low flows at the Glendon site for periods when underflows at the gaging stations on Little Lehigh and Monocacy Creeks are significant. The models were used to route hypothetical releases from Francis E. Walter Reservoir during a low-flow period. The model for the Lehigh River indicated that an added release of 50 ft3/s (cubic feet per second) over a 64-day period during the severe drought in the summer of 1965 would have increased minimum flows for this period at Bethlehem and Glendon by approximately the same amount. A hypothetical release of 200 ft3/s for the period July 20-22, 1965, which is about eight times the actual release in this period, would have been attenuated by about 25 percent when it reached the Bethlehem gage. The synthesized hydrograph for the Bethlehem gage showed such a release would have passed their by July 27. Unresolvable timing errors in the models created an unrealistic hydrographic response for this release at the Trenton gage; but, such a release probably would have passed Trenton by July 29. In order to time the movement of a release wave more accurately than could be done with the developed model, travel times for the wave of an augmentative low-flow release were obtained by field observations and comparisons of gage-height records. The observed leading edge of an abrupt release of 153 ft3/s from Francis E. Walter Reservoir, which ended a 2-day release at a rate of 48 ft3/s, arrived at the gage below the reservoir in 0.5 hour, at White Haven in 3.7 hours, at the mouth of Pohopoco Creek in about 23.1 hours, at Walnutport in 27 hours, at Bethlehem in 39 hours, and at Glendon in 42 hours. This release could not be detected in the record for the Trenton gage. Travel time for an augmentative release in the Lehigh River is dependent upon the pre-release discharge, the relative magnitude of the release, and antecedent rainfall. Relationships are provided for estimating the time of arrival at Walnutport, Bethlehem, and Glendon of the leading edge of waves generated by augmentative releases of 75 to 600 ft3/s. Stage observations on Pohopoco Creek indicated a 2.1-hour travel time between Beltzville Lake and the Lehigh River for the elading edge of a wave produced by a typical augmentative release from this reservoir.