In　subway　stations,　bottlenecks　are　the　narrowed　areas　that　reduce　pedestrian　flows　in　channels.　Because　pedestrians　at　bottlenecks　are　forced　to　dense　together,　bottlenecks　decrease　flow　efficiency　and　pedestrians＇　transfer　comfort　and　may　trigger　serious　crowd　disasters　such　as　trampling.　This　study　used　pedestrian　experiments　to　investigate　the　methods　of　optimizing　pedestrian　traffic　at　bottlenecks　of　subway　stations.　Three　optimization　measures　were　proposed　and　evaluated　by　analyzing　the　characteristics　of　pedestrian　flows,　including　efficiency,　smoothness,　and　security.　In　this　paper,　setting　the　rear　sides　of　the　bottleneck　entrance　as　straight　and　surface　funnel　shapes　is　called　straight　funnel　shape　and　surface　funnel　shape,　respectively.　Setting　a　column　at　a　bottleneck　is　called　the　column　obstacle.　The　results　showed　that　when　efficiency　or　security　come　first,　a　column　on　the　left　is　recommended;　when　comfort　comes　first,　a　concave　funnel　is　recommended;　when　comprehensiveness　is　prioritized,　a　column　on　the　left　is　recommended.　Moreover,　the　larger　the　volume,　the　optimization　is　more　obvious.　Although　many　bottlenecks　cannot　be　prevented　when　subway　stations　are　constructed,　the　proposed　optimization　measures　may　help　ease　their　adverse　effects　by　improving　facility　efficiency,　smoothness,　and　security,　and　by　providing　recommendations　for　designing　and　managing　subway　stations.