Responses on both the state and trait scales of the State-Trait Anxiety (STAI) Inventory were examined under two conditions. The first condition presented a simulated real-life situation containing competitive and evaluative cues without directly suggesting faking and asked subjects to complete the STAI. After an intervening task, the STAI was readministered under standard instructions. The hypothesis that subjects would respond consistently on the A-trait scale, while demonstrating increased levels of A-state was confirmed. Results were interpreted as consistent with the distinction between trait anxiety as a relatively permanent attribute of the individual and state anxiety as a transitory characteristic which fluctuates with the environment. However, the results conflict with traditional notions of social desirability response bias, which imply that "faking good" should lead to consistent changes on both scales. There is little doubt that "faking good" can lead to distortions of scores on affective measures. However, subjects do not necessarily actually "fake good" in situations where they have no definite instructions to fake. (Author/RC)