Methods for assigning tolerance values to different taxa often rely on estimating an optimum for each taxon with respect to a particular environmental gradient.  These estimates of optima (e.g., weighted averages) usually do not take into account other environmental factors that may covary with the factor of interest.  Thus, inferences that are drawn from using these tolerance values can be very uncertain.  Tolerance values may be more useful if they were based on reliable estimates of the underlying cause-effect relationships between environmental conditions and the occurrence of particular taxa.  Here, we introduce a propensity score approach for assigning tolerance values.  Propensity scores provide a way of pre-treating observational data, such that the covariances between the factor of interest and other confounding factors are greatly reduced.  Relationships estimated using this approach are more likely to reflect cause-effect relationships, and optimum values (and associated tolerance values) should more reliably indicate taxon preferences for different environmental conditions.  We apply this approach to assign macroinvertebrate tolerance values for sediment, temperature, and nutrients using data from the EPA Wadeable Streams Survey.