These experiments took place in a field setting in which some of the subjects were unaware of their participation in the experiment. In Experiment 1, 76 students observed a theft and were later asked to identify the perpetrator from a target-absent lineup, receiving either unbiased or biased instructions. One half of the subjects were debriefed prior to the identification procedure. It was found that instructional bias increased the rate of 'don't know' responses for undebriefed subjects, whereas debriefed subjects were unaffected by the type of instructions. In Experiment 2, 63 students who had or had not been debriefed received either biased or unbiased instructions. Cultural and methodological factors were examined to ascertain whether they accounted for the results. The findings demonstrate that witnesses are less susceptible to biased instructions than has been suggested by previous research. (Author abstract modified)