A series of three laboratory experiments examined effects of affirmative action (AA) employment policies, decision set (in terms of accepting or rejecting applicants), and selection ratio (ratio of applicants to available positions) on amount, duration, content, and sequence of information search on job applicants, on personnel selection decisions, and on judgments of applicants. AA was investigated only in conjunction with a concern for hiring qualified applicants and only with respect to applicant gender (not race/ethnicity). The target job was a slightly male sex-typed job. A computerized information board tracked information search. When no decision set was provided, AA increased (decreased) the amount of search on female (male) applicants when considering search on females relative to males and decreased the time spent on males. More females (and fewer males) were hired in the presence than absence of an AA policy unless there was a high selection ratio with no decision set provided. More failure-relevant information about applicants was sought in a reject than accept decision set when there was not an AA policy. Amount of intradimensional search (search by attribute across applicants) was greater in a reject than accept decision set. Decision set also affected ratings of applicant qualifications. The amount and duration of search on applicants who were selected was greater for a low than high selection ratio. The same pattern held for duration per item of information on applicants who were selected if there was an accept decision set or no decision set was provided. Decision makers were more (less) accurate in identifying the best- (least-) qualified applicants when there was a low than high selection ratio. When comparing the number of females hired relative to their proportion in the applicant pool, females were either on equal footing or at an advantage (disadvantage) relative to males in the presence (absence) of an AA policy, depending on decision set and selection ratio. AA attitudes affected information search although not in a consistent manner across experiments. The multitude of effects found highlights the sensitivity of information search in the personnel selection process to person, task, and environmental characteristics.