Nonadherence to inhaled corticosteroid therapy is common and has a negative effect on clinical control, as well as increasing morbidity rates, mortality rates and health care costs. This review was conducted using direct searches, together with the following sources: Medline; HighWire; and the Latin American and Caribbean Health Sciences Literature database. Searches included articles published between 1992 and 2008. The following methods of assessing adherence, listed in ascending order by degree of objectivity, were identified: patient or family reports; clinical judgment; weighing/dispensing of medication, electronic medication monitoring; and (rarely) biochemical analysis. Adherence rates ranged from 30 to 70%. It is recognized that the degree of adherence determined by patient/family reports or by clinical judgment is exaggerated in comparison with that obtained using electronic medication monitors. Physicians should bear in mind that true adherence rates are lower than those reported by patients, and this should be considered in cases of poor clinical control. Weighing the spray quantifies the medication and infers adherence. However, there can be deliberate emptying of inhalers and medication sharing. Pharmacies provide the dates on which the medication was dispensed and refilled. This strategy is valid and should be used in Brazil. The use of electronic medication monitors, which provide the date and time of each triggering of the medication device, although costly, is the most accurate method of assessing adherence. The results obtained with such monitors demonstrate that adherence was lower than expected. Physicians should improve their knowledge on patient adherence and use accurate methods of assessing such adherence.