Observations of hydraulic head response to lagoon leakage into an extremely heterogeneous aquifer allow qualitative identification of spatial patterns in heterogeneity. In a 150-hour leakage experiment over such an aquifer, the fluid mass balance was estimated using a stage-fluctuation model while transient response was observed in downgradient wells and springs. Based on timing with respect to pond leakage, water-level response at different locations within the aquifer may be identified as either within or outside the influence of preferred-flow paths (PFPs). Flows at springs 150-200 m from the pond increased by 20-60% above baseline flows, showing initial increase within ca. 40 hours, Synoptic water-level rise patterns observed during the test describe enlargement of a recharge mound from beneath the pond and suggest the spatial pattern of PFPs to > 150 m distant. An analytical model assuming homogeneity (Hantush, 1967), applied to water-level response for wells > 30 m from the pond, overestimated K by > 1 order of magnitude in comparison to slug-test estimates. This discrepancy is attributed to the inferred presence of linearly continuous heterogeneities causing preferential flow. Large-scale hydraulic tests such as this sample the spatial structure for conductive pathways and may be of utility in parameterization of numerical flow models. In contrast, slug-test results give local parameter estimates, but more limited information on PFP distribution.