Purpose: To compare pay gap estimates across 3 different national survey data sets for people with disabilities relative to those without disabilities when pay is measured as wage and salary alone versus a (total compensation) definition that includes an estimate of the value of benefits. Method: Estimates of the cost to the employers of employee benefits at the occupational level from an employer survey data set are matched to individual-level data in each of the 3 data sets. Multiple regression techniques are applied to estimate wage and salary and total compensation gaps between full-time men with and without disabilities. Results: For full-time working men with disabilities (relative to those without disabilities), there is a consistently larger percentage wage and salary gap than percentage total compensation gap and breadth of the definition of pay affects the size of any estimated pay gap. In addition, there are differences in the estimated pay gaps depending on data source and disability measure. Conclusions: Results obtained from a single data set or definition of key variables may not be broadly generalizable. Studies containing such limitations should be interpreted cautiously. Our research further suggests employers looking to hire persons with disabilities or those offering employment placement services should put substantial weight on the non-base pay component of the total compensation package.