The use of inverse one-dimensional (1-D) analytical methods for estimating vertical stream-aquifer exchange flux is now commonplace. However, the application of such simple models can lead to significant errors in estimates of vertical exchange flux where the model assumptions are violated in real systems. An idea that is gaining acceptance in the literature is that the presence of nonvertical flow is such a violation. However, it is shown here that nonvertical flow by itself will not necessarily lead to errors in vertical flux estimation but rather that significant errors can stem from nonuniform (convergent/divergent) flow fields and/or hydrodynamic dispersion even within uniform flow fields. Nonuniform flow may also be expected, in some cases, to create discrepancies between flux estimates made on the basis of vertical head gradient measurements and those made using 1-D analytical heat tracer methods. Significant differences are observed in the estimates of heat-derived fluxes obtained by the amplitude ratio and phase-shift time-series methods when convergent and divergent flows are apparent. Such differences may potentially be used to infer that convergent or divergent flow is occurring and that a 1-D analysis is inappropriate.