Nondirectiveness is considered an essential part of genetic counseling, yet there is no generally accepted definition nor data documenting its impact on counselees. This study is an empirical investigation of directiveness, using ratings from transcripts of consultations and comparing these with counselor-reported and counselee-reported directiveness. Rated directiveness was defined as advice, expressed views about or selective reinforcement of counselees' behavior, thoughts, or emotions (advice, evaluation, and reinforcement). Analysis of 131 transcripts revealed a mean of 5.8 advice statements per consultation, 5.8 evaluative statements, and 1.7 reinforcing statements. When asked to describe their counseling style, none of the 11 counselors rated it as "not at all" directive. Half the counselees who faced a decision felt steered by the counselor. Items of rated directiveness showed satisfactory interrater reliability (kappa = .63). Factor analysis revealed that they formed one factor (eigenvalue 1.72). There were no associations either between counselor-reported, counselee-reported, and rated directiveness or between these measures and counselee anxiety and concern, satisfaction with information, or the meeting of counselees' expectations. Rated directiveness was the only measure to be associated with other process measures of the consultation, being associated with longer consultations, more blocks of speech, more social and emotional issues being raised, and fewer concerns being followed up. Advice was more likely to be given to counselees of lower socioeconomic status and to counselees judged by counselors to be highly concerned. Evaluative statements were more likely to be made by counselors who had received counseling training. These results show that genetic counseling was not characterized--by counselors, counselees, or a standardized rating scale--as uniformly nondirective.