Few have clarified what exercise frequencies and patterns (e.g. alone or with others) are effective for preventing depression in older adults. We examined the relationship between total frequency and/or pattern of exercise and the risk of depression after two years in older Japanese adults. We used a sub-sample of the Japan Gerontological Evaluation Study (JAGES) performed in 2011 and 2013. The sample for analysis was 1,422 adults aged 65 years or older without depression and low physical strength in 2011. All variables were assessed with a questionnaire including the geriatric depression screening scale (GDS-15). Binomial logistic regression analysis was used to examine the relationships between exercise in 2011 and depression in 2013 (0 = non-depression, 1 = depression). The adjusted odds ratio (OR) for later depression was 0.52 (95% confidence intervals: 0.33-0.81) for exercise two or more times a week compared to non-exercisers. The OR for exercisers who exercise with others even a little (Ewo) was 0.53 (0.34-0.84) compared to non-exercisers. Among combinations of frequency and pattern, the OR for Ewo who exercise two or more times a week was 0.40 (0.24-0.68) compared to non-exercisers. Exercising at least twice a week and/or with others may be useful in preventing depression in older adults.