For　more　accurate　estimation　of　urban　traffic　emissions,　it　is　necessary　to　study　the　emission　estimates　under　different　traffic　conditions.　In　this　paper,　a　novel　emission　estimation　method　on　the　basis　of　levels-of-service　(LOS)　is　presented　to　estimate　city　emissions　under　different　traffic　conditions.　The　primary　contributions　of　our　work　are　a　novel　set　of　velocity-acceleration　features　to　distinguish　different　urban　levels-of-service,　and　an　emission　estimation　method　on　the　basis　of　levels-of-service　classification.　The　processes　of　our　method　includes　1)　velocity-acceleration　feature　extraction,　2)　classification　by　learning　algorithms,　3)　levels-of-service　construction　and　4)　emission　estimation　by　existed　emission　models.　For　better　illustration,　a　LOS　A-F　classification　system　was　created　for　public　transits　in　Beijing　and　the　practical　emission　factors　were　calculated　based　on　LOS.　Our　levels-of-service　classification　was　evaluate　by　using　decision　tree　C4.5　algorithm　for　over　14　hours　of　real　transportation　data.　The　results　show　that　accuracy　rate　of　our　classification　is　greater　than　98%.　The　on-road　emission　inventory　under　LOSs　was　put　forward　to　benefit　further　analysis　and　evaluation　of　emissions　as　a　reference.　©　2016　IEEE.