Numerical simulations were used to examine the effects of heterogeneity in hydraulic conductivity (K) and intrinsic biodegradation rate on the accuracy of contaminant plume‐scale biodegradation rates obtained from field data. The simulations were based on a steady‐state BTEX contaminant plume undergoing biodegradation under sulfate‐reducing conditions, with the electron acceptor in excess. Biomass was either uniform or correlated with K to model spatially variable intrinsic biodegradation rates. A hydraulic conductivity data set from an alluvial aquifer was used to generate three sets of 10 realizations with different degrees of heterogeneity, and contaminant transport with biodegradation was simulated with BIOMOC. Biodegradation rates were calculated from the steady‐state contaminant plumes using decrease in concentration with distance downgradient and a single flow velocity estimate, as is commonly done in site characterization to support the interpretation of natural attenuation. The observed rates were found to underestimate the actual rate specified in the heterogeneous model in all cases. The discrepancy between the observed rate and the “true” rate depended on the ground water flow velocity estimate, and increased with increasing heterogeneity in the aquifer. For a lognormal K distribution with variance of 0.46, the estimate was no more than a factor of 1.4 slower than the true rate. For an aquifer with 20% silt/clay lenses, the rate estimate was as much as nine times slower than the true rate. Homogeneous‐permeability, uniform‐degradation rate simulations were used to generate predictions of remediation time with the rates estimated from the heterogeneous models. The homogeneous models generally overestimated the extent of remediation or underestimated remediation time, due to delayed degradation of contaminants in the low‐K areas. Results suggest that aquifer characterization for natural attenuation at contaminated sites should include assessment of the presence and extent of, and contaminant concentrations in, low‐permeability areas of an aquifer.