Medication adherence is a well-recognized problem in the management of patients with chronic diseases needing a long-term pharmacotherapy. While fighting against non-adherence, an important question frequently arises, i.e., how much adherence is enough to obtain the full treatment benefits? Most studies having attempted to answer this question have used large pharmacy records and health care databases to quantify the percentage of days covered by the prescribed treatment and hence deduce a threshold below which there appears to be fewer benefits from therapy. In the present paper, the use of data obtained using electronic monitoring of adherence is discussed as another means to assess adherence thresholds with a particular emphasis on hypertension. The data show that even with the use of electronic monitoring of adherence, which provides a comprehensive dosing history, it is extremely difficult to define an adherence threshold in hypertension. This is due to many factors that need to be taken into account, including not only the pattern of patients' adherence and their clinical and environmental characteristics, but also the pharmacological characteristics of the prescribed drugs, the severity of the disease and many others. To determine adherence cut-offs more precisely, specific protocols should be designed to answer the question in various clinical conditions. These protocols should be conducted in well-defined patients' groups, they should use the most reliable methods to measure adherence providing if possible a detailed dosing history perhaps combined with drugs levels in blood or urine. These studies should also choose the best methods to measure clinical endpoints, such as ambulatory blood pressure monitoring or home blood pressure in the case of hypertension. One important aspect is that datasets should be solid and large enough to be able to analyze adherence data as a continuous variable using newly developed mathematical models including new metrics catching the complexity of adherence. The rapid development of new technologies like devices, connectivity, and analytics, will probably provide new solutions to improve our ability to define valid and clinically useful adherence thresholds in various therapeutic areas.