To assess the acquisition and retention of screening ophthalmic clinical skills over 3 years of medical school. Observational, longitudinal, multiple skills measures. All 96 students enrolled in a single graduating class at a public medical school in California. Immediately after the second-year ophthalmic clinical skills course, all students were evaluated by their preceptors and self-rated for competence in defined skills. Follow-up assessments were done during 2 required third-year clerkships without additional formal ophthalmic clinical exposure. Three complete history and physical examination chart notes routinely submitted for course grading in third-year clerkships were selected randomly for each student by clerkship directors in family medicine and internal medicine, masked for identity, and then scored for appropriateness. Funduscopic skills were assessed objectively with a simulator in the third-year Clinical Performance Examination. During a fourth-year 4-day ophthalmology clerkship, students were trained and reassessed with the same simulator. Just before graduation, a self-assessment questionnaire was administered to the entire class. Twelve skills were assessed: ability to evaluate visual acuity (VA); pupils; extraocular muscles; confrontation fields; lids; cornea; conjunctiva/sclera; anterior chamber depth; and, funduscopically, the disc, macula, vessels, and retina. Faculty rated 88% to 90% of students as able to assess acuity, pupils, ductions, and fields, and 72% to 82% as able to visualize various parts of the fundus. Seventy-six percent of students felt comfortable after funduscopic training. In 364 analyzed chart notes, one VA was measured, and pupils were examined in 66% of notes, ductions in 54%, and fields in 3%. Only 11% of notes documented attempted funduscopy; <2% suggested actual visualization. In the Clinical Performance Examination, 32% of students accurately described some aspect of the disc, with an improvement to 84% of 38 students retested after brief ophthalmology training in year 4. Of 54 (56%) respondents to the exit questionnaire, 59% felt comfortable visualizing some aspect of the fundus. There is worrisome erosion of students' acquired screening skills across the third-year clerkships. Skill reinforcement in the fourth year yielded improved performance. Attention must be directed to reinforcing basic ophthalmology skills training within medical school curricula to assure competence of graduates.