Next　generation　sequencing　technology　has　led　to　the　development　of　methods　for　the　detection　of　novel　sequence　insertions　(nsINS).　Multiple　signatures　from　short　reads　are　usually　extracted　to　improve　nsINS　detection　performance.　However,　characterization　of　nsINSs　larger　than　the　mean　insert　size　is　still　challenging.　This　article　presents　a　new　method,　ERINS,　to　detect　nsINS　contents　and　genotypes　of　full　spectrum　range　size.　It　integrates　the　features　of　structural　variations　and　mapping　states　of　split　reads　to　find　nsINS　breakpoints,　and　then　adopts　a　left-most　mapping　strategy　to　infer　nsINS　content　by　iteratively　extending　the　standard　reference　at　each　breakpoint.　Finally,　it　realigns　all　reads　to　the　extended　reference　and　infers　nsINS　genotypes　through　statistical　testing　on　read　counts.　We　test　and　validate　the　performance　of　ERINS　on　simulation　and　real　sequencing　datasets.　The　simulation　experimental　results　demonstrate　that　it　outperforms　several　peer　methods　with　respect　to　sensitivity　and　precision.　The　real　data　application　indicates　that　ERINS　obtains　high　consistent　results　with　those　of　previously　reported　and　detects　nsINSs　over　200　base　pairs　that　many　other　methods　fail.　In　conclusion,　ERINS　can　be　used　as　a　supplement　to　existing　tools　and　will　become　a　routine　approach　for　characterizing　nsINSs.