Metered-dose inhalers are often used incorrectly by patients with chronic airflow obstruction, and there is a lack of controlled studies designed to evaluate methods to teach the correct use of these devices. Therefore, we screened 100 consecutive stable outpatients for correct or incorrect inhaler use and then conducted a randomized trial of two methods to teach correct use. Patients were classified as correct or incorrect users with a modified metered-dose inhaler containing a thermistor that detected inspiration, inhaler activation, and the duration of breath-holding. Patients were classified as incorrect users if they failed to: (1) activate the inhaler once during inspiration, and/or (2) hold their breath at the end of inspiration. There were 38 correct users and 62 incorrect users. Incorrect users were randomized to one of two teaching protocols: (1) standardized verbal instruction alone, or (2) standardized verbal instruction plus an automatic visual signal during inspiration. Incorrect users were restudied 6 to 10 wk later to reassess technique. Both treatment protocols were equally effective in converting incorrect user. However, the verbal instruction alone required significantly less time than the instruction with the mechanical aid. For all subjects, the proportion using correct technique declined over time, particularly for incorrect users. We also examined a series of patient characteristics, obtained by questionnaire and spirometry, to determine whether they could be used in the clinical setting to identify incorrect users. By discriminant analysis, a group of four variables predicted correct metered-dose inhaler use: bronchodilator responsiveness, a history of additional about proper technique, verbal knowledge of the correct inhaler maneuvers, and the patient's perception of whether it is important to use an inhaler.(ABSTRACT TRUNCATED AT 250 WORDS)