The Bureau of Labor Statistics (BLS) Current Employment Statistics (CES) program currently collects earnings and hours data for production and nonsupervisory workers. A review of this program concluded that user needs would be better served if CES collected earnings and hours for all employees. In a pilot test we compared two approaches to measuring all-employee earnings: (1) gross payroll for the pay period that includes the 12th of the month (payroll includes overtime, but excludes irregular bonuses, commissions, or other nonwage payments) plus a measure of nonwage payments for the entire previous month, and (2) total earnings for the entire previous month (total earnings includes all payments made to employees). Two groups of 500 establishments each were asked to report earnings and the associated hours using concept (1) or concept (2). We found that response rates for gross payroll and total earnings were comparable after controlling for observable differences between the groups.The response rate for nonwage payments was lower than for either gross payroll and total earnings. To assess data quality, we matched reported earnings data to those collected from the Unemployment Insurance (UI) records. Again, there was no difference between the two concepts.