Traditional　transaction-based　and　run-on-a-stand-alone　clustering　algorithms　cannot　meet　the　demand　for　knowledge　discovery　in　massive　trajectories　data.　This　paper　presents　a　MapReduce　-based　distributed　parallel　algorithm　to　extract　the　hot　path　from　the　taxi　track.　Compared　with　the　hot-zone　extracting　algorithms　based　on　traditional　density-based　clustering　algorithms,　the　developed　algorithm　is　concise　and　easy　to　implement.　Experiments　on　the　actual　taxi　trajectory　data　sets　show　that　the　algorithm　performs　well　on　large-scale　distributed　data　sets.