Occupants＇　behaviour　has　a　significant　impact　on　the　energy　performance　of　buildings.　A　good　understanding　of　how　occupants　use　a　building　provides　a　possibility　of　promoting　the　building＇s　energy　efficiency　through　changing　occupant　behaviour.　Building　simulation　has　been　adopted　as　a　useful　method　by　building　engineers　for　quantifying　the　effects　of　changing　occupant　behaviour　on　the　building＇s　energy　consumption　and　indoor　environment.　However,　due　to　the　lack　of　real　measured　data　with　respect　to　how　occupants　use　the　building,　such　simulation　work　has　relied　on　assumed　behavioural　patterns,　which　significantly　reduces　the　reliability　of　the　predicted　results.　This　paper　describes　a　longitudinal　study　monitoring　occupants＇　heating,　window　opening　and　cooling　behaviour　in　an　office　building　throughout　summer,　transitional　and　winter　periods.　These　behavioural　data　were　then　used　to　drive　dynamic　building　performance　simulation　to　predict　the　energy　saving　potential　of　changing　behaviour.　Comparison　with　predicted　results　by　assumed　behavioural　patterns　reflected　that　improperly　assumed　behavioural　patterns　may　either　overestimate　or　underestimate　the　energy　saving　potential　of　changing　behaviour,　especially　for　unextreme　behaviours.