In-vehicle　thermal　environment　is　important　to　driving　behaviours.　The　behaviours　are　always　related　to　thermal　comfort,　fatigue　and　information　reading,　which　are　sensitive　to　the　interior　design.　However,　previous　studies　have　merely　analysed　the　impact　of　in-vehicle　temperature　and　in-vehicle　design　on　driving　behaviours　separately.　There　is　still　a　gap　to　understand　how　the　two　affect　driving　behaviours　simultaneously.　Therefore,　this　study　aims　to　investigate　the　driving　behaviours　(e.g.,　eye　movement　characteristics　and　completion　time　of　operation　tasks)　subject　to　different　combinations　of　temperature　and　interface　layouts.　The　study　was　conducted　based　on　a　laboratory　experiment　among　53　subjects,　extreme　learning　machine　(ELM)　model　and　entropy　weight　method.　The　subjects　were　inquired　to　conduct　two　tasks　including　navigation　and　music　control　for　analysing　the　variation　of　fixation　duration,　fixation　times　and　completion　time.　The　temperature　conditions　consisted　of　six　levels　ranging　from　22　to　32　°C　and　there　were　two　types　of　interface　layouts　(checkerboard　type　and　linear　type).　The　results　show　a　non-linear　relationship　between　completion　time　and　the　factors　of　temperature,　interface　layout　and　operation　task,　and　the　co-impacts　of　these　factors　were　more　complex.　Furthermore,　the　analysis　indicates　that　the　most　suitable　temperatures　for　navigation　and　music　control　were　22　and　24　°C.　Moreover,　the　ELM　model　outperformed　the　back　propagation　neural　network　(BPNN)　model　in　predicting　the　relationship　between　visual　ergonomics　and　operation　efficiency　with　the　intervention　of　ambient　temperature.　In　addition,　entropy　weights　imply　that　both　fixation　duration　and　completion　time　were　more　sensitive　to　in-vehicle　temperature　than　fixation　times.　Overall,　beyond　previous　studies　on　the　impact　of　in-vehicle　design　on　driving　behaviours,　this　study　considers　the　additional　heat-induced　impacts　on　driving　behaviours,　which　is　meaningful　to　provide　a　reference　for　in-vehicle　environment　design.　©　2022