Thymoma patients who do not have history of myasthenia gravis sometimes develop the disease following thymectomy. The aim of the study was to clarify the risk factors for developing post-thymectomy myasthenia gravis in thymoma patients. We performed a retrospective analysis of patients, without preoperative myasthenia gravis, who underwent thymectomy for thymoma between 1991 and 2011 at our institution. Patient characteristics and preoperative acetylcholine receptor antibody levels were recorded, and the presence or absence of post-thymectomy myasthenia gravis was noted. Among a total of 229 patients, 123 had no myasthenia gravis history at the time of operation. A positive result for acetylcholine receptor antibody, >0.2 nmol/l, was noted in 33 patients; of these, 10 developed post-thymectomy myasthenia gravis and 23 did not. The 90 patients with negative acetylcholine receptor antibody levels did not develop the disease. The mean day of myasthenia gravis onset was postoperative day 988 (range, postoperative day 3–2859). Patients with positive acetylcholine receptor antibody levels were significantly more likely to develop post-thymectomy myasthenia gravis than patients with negative levels (P < 0.0001). Other factors, including patient age, sex, respiratory function, surgical approach, tumour size, and Masaoka classification, were not associated with onset of the disease. Comparing patients who did and did not develop post-thymectomy myasthenia gravis, there were significant differences in the acetylcholine receptor antibody titres (12.2 ± 4.5 nmol/l vs 4.7 ± 1.1 nmol/l; P = 0.0039), the World Health Organization pathological classification (ratio of type B1, B2 and B3 to all, 80% vs 18.8%; P = 0.0048), and the rate of tumour recurrence (30% vs 0%; P = 0.0220). A positive preoperative acetylcholine receptor antibody level is a risk factor for developing post-thymectomy myasthenia gravis in thymoma patients. Patients with higher acetylcholine receptor antibody titres are more likely to develop the condition. All authors have declared no conflicts of interest.