Reports of metered inaccuracy in 'therapeutic' ultrasound unit output have been made since 1962, but have tacitly been accepted, perhaps due to the universal lack of appropriate testing facilities. Factors of treatment selection subject to instrumental error include duration of application, operating frequency, intensity, pulsed output. Metered errors in space-averaged intensity are common and are the most difficult to detect without specialised equipment which is rarely available. Couplants, essential for ultrasound transmission, can be a source of acoustic power loss if incorrectly used. Beam profiles demonstrate the rapid spatial variations in the near (Fresnel) zone, necessitating soundhead movement during treatment. Ultrasound physics must be understood, but professional integrity should demand better output testing facilities and equipment for clinical treatment.