The　product　manufactured　by　selective　laser　sintering　(SLS),　usually　applied　in　high　value　engineering　products　including　an　airplane,　medical　support　and　so　on,　and　needs　a　careful　assessment　and　analysis　of　the　defects.　Defects　induced　during　the　3D　printing　metals　by　SLS　process　are　difficult　to　be　predicted　and　eliminated.　The　aim　of　this　study　is　to　observe　and　obtain　the　statistics　of　defects　so　as　to　utilize　the　numerical　predictions　to　quantify　the　variation　in　the　mechanical　properties　by　considering　statistical　nature　of　the　defects.　Specifically　one　cellular　structure　is　fabricated　several　times　as　variation　of　building　directions　in　two　different　ways.　After　manufacturing,　the　Computed　Tomography　(CT)　technology　is　used　to　observe　the　microstructure　in　order　to　evaluate　the　performance　of　printed　structure.　In　the　raw　material　of　managing　steel,　the　geometrical　defects　are　characterized　and　categorized　into　three　main　groups,　i.e.　notch,　kink　and　hole.　It　is　found　that　the　directions　of　above　defects　are　put　along　with　the　direction　of　printing.　By　reconstruction　of　CT　images,　the　3D　model　of　printed　structure　is　built　and　the　parametrization　of　those　geometrical　defects　and　statistics　are　done.　It　is　also　found　that　when　the　printing　angle　is　set　at　45°,　the　kink　is　of　most　probable　to　be　formed　while　the　notch　is　likely　to　be　formed　at　0°.　Based　on　those　statistics　data,　microstructures　including　defects　are　built　with　random　parameters.　The　macroscopic　properties　are　analyzed　to　obtain　the　variation　from　mean　value　in　macroscopic　stiffness,　permeability　and　micro　stress.　Variation　turned　out　to　be　huge　of　micro　stress　but　small　in　macro　stiffness.　The　numerical　result　is　validated　via　compression　test　of　macroscopic　stiffness　for　printed　specimens.　This　study　can　contribute　to　validation　of　numerical　models　and　obtained　statistics　is　also　applicable　in　other　3D　printed　structure.　©　2018　Journal　of　Mechanical　Engineering.