Anti-saccade performance, with strong contributions from frontal brain regions, reportedly deteriorates with age and maintenance of neck flexion and is known to cause brain activation. We investigated the effects of anti-saccade training on eye movement performance and frontal activity, and synergistic effects of training with neck flexion in the elderly. Thirty elderly individuals were divided into three equal groups: training group at neck resting position (NRT); training group at 20° neck flexion position (NFT); and untrained group. NRT and NFT performed approximately 200 anti-saccades (a block of 10-12 anti-saccades for 30 s × 20 blocks) per day over 3 weeks. Before and after training, horizontal eye movement, presaccadic potentials, and oxygenated hemoglobin concentration (oxy-Hb) in the prefrontal cortex during anti-saccades were tested in neck resting and 20° neck flexion conditions. In NRT and NFT, reaction time (-50 ms), percentage of erroneous saccades (-24%), and period between peak of presaccadic negativity and onset of spike potential (-16 ms) were significantly decreased through training. Only in NFT, after training, slight shortening of reaction time associated with neck flexion was recognized (-10 ms), and peak amplitude of presaccadic negativity was increased in both test neck conditions. Oxy-Hb was not significantly affected by trainings and test neck conditions. We demonstrated that in the elderly, anti-saccade training with both neck postures improved performance and facilitated related neural pathways. Moreover, training with neck flexion showed small but synergistic effects on performance and frontal activity. However, these trainings would be insufficient for elderly individuals to automatically control anti-saccade.