With the shift of a large proportion of cancer chemotherapy recipients to ambulatory care, the role of hospital pharmacists has changed, and their provision of information is essential care for cancer patients. There is little research on pharmacist–patient relations, particularly about pharmacist counselling, in Japan. To meet patients’ needs, pharmacist counselling should be optimized. Here, breast cancer patients’ preferences for pharmacist counselling were assessed using a discrete choice experiment. Bayesian nonlinear optimal methodology was employed to obtain six attributes (attitude of pharmacist, quality of information, explanation of side effects, frequency of pharmacist counselling before starting chemotherapy, cost of pharmacist counselling, and follow-up with the pharmacist after starting chemotherapy) of two to three levels each. The attributes and levels were used to create 12 hypothetical scenarios that were divided into two questionnaires of six choice sets each. Two hundred eighty participants were randomly assigned to complete one of these questionnaires (blocks). Attributes were analyzed by conditional logit model to determine significant predictors of patient preferences. The responses of 278 patients to 1667 scenarios were analyzed. Attitude of pharmacist, quality of information, cost of pharmacist counselling, and follow-up with the pharmacist after starting chemotherapy were significant predictors of patient preferences, with quality of information receiving the highest priority. Thus patients receiving pharmacist counselling before starting chemotherapy prefer to interact with a pharmacist with a friendly, interested attitude who provides individualized information. Further research is needed to elucidate the information that Japanese patients consider most important and to enhance pharmacist–patient communication.