Aim: The optimal treatment of ocular myasthenia gravis (OMG) remains unknown. The authors evaluated the efficacy of prednisone and pyridostigmine in reducing diplopia, ocular motor dysfunction, and ptosis in patients with OMG. Methods: Review of records from a clinical database from one neuro-ophthalmology service of patients presenting with OMG between 1990 and 2002, excluding those who developed generalised MG within the first month after diagnosis. Institutional review board approval was obtained for this study. Participants/interventions: Non-randomised, unmasked, therapy was given. 55 patients with diplopia in primary or downward gaze and clinically demonstrable extraocular muscle dysfunction received prednisone. 34 patients who had contraindications to steroids or who refused treatment with prednisone received pyridostigmine only. Over 5 days the daily prednisone dose was increased to 50-60 mg and then gradually reduced to 10 mg, followed by further reduction as tolerated. The pyridostigmine dose was begun at 180 mg daily and increased as tolerated. Main outcome measures: Follow up evaluations, performed at 1, 3-6, 12, and 24 months, detailed the frequency of ptosis and diplopia and the amount of ocular motor deviation in primary and downward gaze. Results: The prednisone and pyridostigmine groups were similar for age, sex, acetylcholine receptor antibody level, prism cover test results for primary and downward gaze, diplopia in primary and downward gaze, and unilateral ptosis. Bilateral ptosis was present in 32.4% of the pyridostigmine group and 10.9% of the prednisone group (p = 0.02). The prednisone group showed resolution in primary gaze diplopia, downgaze diplopia, unilateral ptosis, and bilateral ptosis in 73.5%, 75.5%, 85.7%, and 98%, respectively at 1 month. The benefit persisted at 3-6, 12, and 24 months except for the bilateral ptosis. The pyridostigmine group showed resolution in primary gaze diplopia, downgaze diplopia, unilateral ptosis, and bilateral ptosis in 6.9%, 17.2%, 50%, and 76.7% of patients after 1 month of treatment. The prism cover results improved (p = 0.003) in the prednisone group only. In the prednisone group, four patients had no response to therapy. Among the 51 prednisone responsive patients, there were 33 recurrences in 26 patients. 12 patients, all prednisone treated, had remissions. Except for three patients who developed diabetes, no patient developed a clinically significant systemic corticosteroid complication. Conclusion: These results suggest that 50-60 mg daily prednisone followed by lower doses (10 mg or less) has the benefit of resolving ptosis and diplopia that lasts for at least 2 years in approximately 70% of patients.