This questionnaire study involved the cooperation of home helpers and community pharmacists to improve medication adherence of in-home patients. The survey form contained 18 items that are factors for adherence and a visual analog scale to record the predicted medication adherence (predicted adherence). Assisted by the 14 offices of the home-visit helper care system located in Niiza, Saitama prefecture, 140 in-home patients were surveyed. For 21 of the 140 patients, a pharmacist was able to measure medication adherence by counting the number of pills remaining in the patient's home (true adherence). Factors influencing the predicted and true adherence were analyzed by multivariate analysis and found to be different. Home helpers predicted the medication adherence based on a patient's everyday life such as "irregular meal", "storing up drugs" and "one dose package". On the other hand true adherence was influenced by "urging to take the medicine by the home helper" and "pharmacist-visit". Furthermore, the medication adherence of patients who were not visited by a pharmacist and urged to take medicine by the home helper was low when the (1) age was high, (2) care or support level required was low, and (3) self-control of dosing was suspected. Therefore the home helper should encourage the patient to take the medicine and when the pharmacist is informed by the home helper about patients who fit the above (1)-(3), the pharmacist should visit the patient's home.