In　recent　years,　the　rapid　development　of　industrial　technology　has　been　accompanied　by　serious　environmental　pollution.　In　the　face　of　numerous　environmental　pollution　problems,　particulate　matter　(PM2.5)　which　has　received　special　attention　is　rich　in　a　large　amount　of　toxic　and　harmful　substances.　Furthermore,　PM2.5　has　a　long　residence　time　in　the　atmosphere　and　a　long　transport　distance,　so　analyzing　PM2.5　distributions　is　an　important　issue　for　air　quality　prediction.　Therefore,　this　paper　proposes　a　method　based　on　convolutional　neural　networks　(CNN)　and　long　short-term　memory　(LSTM)　networks　to　analyze　the　spatial-temporal　characteristics　of　PM2.5　distributions　for　predicting　air　quality　in　multiple　cities.　In　experiments,　the　records　of　environmental　factors　in　China　were　collected　and　analyzed,　and　three　accuracy　metrics　(i.e.,　mean　absolute　error　(MAE),　root　mean　square　error　(RMSE),　and　mean　absolute　percentage　error　(MAPE))　were　used　to　evaluate　the　performance　of　the　proposed　method　in　this　paper.　For　the　evaluation　of　the　proposed　method,　the　performance　of　the　proposed　method　was　compared　with　other　machine　learning　methods.　The　practical　experimental　results　show　that　the　MAE,　RMSE,　and　MAPE　of　the　proposed　method　are　lower　than　other　machine　learning　methods.　The　main　contribution　of　this　paper　is　to　propose　a　deep　multilayer　neural　network　that　combines　the　advantages　of　CNN　and　LSTM　for　accurately　predicting　air　quality　in　multiple　cities.　©　2019,　Springer　Nature　Switzerland　AG.