Industrial　green　development　(IGD)　is　fundamental　for　curbing　serious　pollution　and　over-consumption　of　natural　resources　caused　by　industrial　growth,　while　quantification　of　the　IGD　level　has　become　instrumental　in　guiding　further　progress.　However,　three　existing　research　gaps　prevent　a　comprehensive　and　accurate　evaluation　of　the　IGD　level.　These　gaps　include　the　ignorance　of　the　underlying　conditions,　lack　of　spatio-temporal　comparability　in　the　IGD　indicator　system,　and　absence　of　more　reasonable　assumptions　within　the　technique　for　order　preference　by　similarity　to　an　ideal　solution　(TOPSIS)　and　slack-based　measure　(SBM)　methods.　The　main　goals　of　this　study　are　to　propose　an　improved　evaluation　framework　to　fill　these　gaps　and　to　provide　new　knowledge　by　analyzing　the　IGD　status,　weaknesses,　and　spatio-temporal　differences.　With　the　collected　panel　data　across　30　provinces　in　China　from　2006　to　2015,　the　following　findings　were　revealed　based　on　the　proposed　evaluation　framework:　1)　both　the　performance　and　quality　of　IGD　increased　but　their　growth　rates　had　significantly　declined　in　the　later　years;　2)　pollution　abatement　had　the　maximum　improvement,　followed　by　low-carbon　production,　resource　reduction,　and　economic　operation;　3)　coastal　provinces　had　higher　performance　and　quality　of　IGD　than　inland　provinces,　especially　for　low-carbon　production　and　pure　technical　efficiency;　4)　provinces　with　higher　IGD　performance　usually　had　higher　pure　technical　efficiency,　but　not　scale　efficiency;　5)　the　traditional　evaluation　framework　either　significantly　overestimated　or　underestimated　the　IGD　efforts　in　some　provinces.　Our　findings　highlight　the　importance　of　differentiated　policies　between　coastal　provinces　and　inland　provinces　or　between　high-IGD-performance/quality　provinces　and　low-IGD-performance/quality　provinces,　as　well　as　call　for　effective　measures　in　response　to　slowing　growth.　Moreover,　our　methodology　can　comprehensively　and　accurately　evaluate　the　IGD　level　by　applying　an　extensive　process　considering　underlying　conditions/inputs,　a　spatio-temporal　indicator　system,　and　new　integrated　methods.　©　2020　Elsevier　Ltd