Medicines work better if taken, which must be true of migraine prophylaxis. There is evidence that compliance with regular medication can be badly deficient. To assess how serious the problem might be in routine migraine management, we undertook a covert observational 2-month survey in a specialist headache clinic using objective measures of compliance. Subjects were 38 patients needing prophylaxis with medication prescribed once (od), twice (bd), or three times daily (tds). Medication was dispensed, unknown to them, in Medication Event Monitoring Systems (MEMS) to record openings in real time. Number, timing, and pattern of actual openings were compared with what was expected. Compliance rates averaged 66%, although returned pill counts indicated 91%. A substantial and significant difference was shown between od and bd or tds regimens. Measures of dosing interval—used-on-schedule rate and therapeutic coverage—averaged between 44% and 71%. Once-daily treatment was associated with a used-on-schedule rate more than double those of multiple daily dosing, but still only 66%. We conclude that routine use of drug prophylaxis in migraine may be so seriously undermined by poor compliance that it has little chance of efficacy. Returned-pill counting is inadequate for compliance assessment.