The proliferation of statewide integrated data systems linking data across state agencies provides opportunities to examine the efficacy of state programs. Since all states run Unemployment Insurance (UI) compensation programs, wage data collected from employers can be a valuable outcome measure. However, research using a state’s UI wage records can introduce biased results if an analyst is not careful. This paper examines the different methods for computing annual wages using UI wage records and attempts to identify the method that yields the lowest bias.