Intraocular pressure was measured in 132 consecutive patients over the age of 20 in a general practice with two general practitioners (GP). Tono-Pen tonometry was compared to Goldmann applantation tonometry in a masked, randomized design. Also, agreement between Tono-Pen measurements by the GPs and the ophthalmologist was estimated. Both mean difference between Tono-Pen and Goldmann applantation measurements and mean differences between Tono-Pen readings made by GPs and ophthalmologist were less than 1 mmHg. Ninety-five percent limits of agreement between the two methods were approximately -4.5 mmHg and 5.5 mmHg, and between observers approximately -4 mmHg and 5 mmHg. The Tono-Pen was agreeable both to patients and physicians. The GPs made as accurate measurements as the ophthalmologist. The wide limits of agreement are clinically much more important than the small differences between means. Random variation was much greater than systematic variation. Wide limits of agreement will apply to all known methods of clinical tonometry.