Corona　Virus　Disease　2019　(COVID-19),　caused　by　the　severe　acute　respiratory　syndrome　coronavirus　2　(SARSCoV-2),　poses　a　serious　threat　to　human　health　and　life　safety.　How　to　effectively　prevent　and　treat　COVID-19　is　crucial.　In　this　study,　we　used　the　inhibitors　of　nonstructural　protein　Nsp14　of　SARS-CoV-2　to　perform　the　quantitative　structure　activity　relationship　(QSAR)　modelling　for　the　first　time.　Based　on　different　dataset　division　strategies,　we　selected　partial　least　square　(PLS)　and　multiple　linear　regression　(MLR)　methods　to　develop　easily　interpretable　and　reproducible　QSAR　models　with　2D　molecular　descriptors.　All　models　complied　with　the　strict　QSAR　validation　principles　of　OECD　and　internationally　recognized　validation　metrics.　The　best　model　contained　two　molecular　descriptors　with　the　following　statistical　parameters:　R2　=　0.7796,　Q2LOO=　0.7373,　R2test　=　0.8539　and　CCCtest　=　0.9073.　Obviously,　the　model　exhibited　good　prediction　performance　and　can　be　used　for　quickly　predicting　the　inhibitory　activity　of　unknown　compounds　against　Nsp14.　Mechanistic　interpretation　identified　the　detailed　relationship　between　molecular　structure　information　and　inhibitory　activity.　The　best　QSAR　model　was　used　to　predict　the　inhibitory　activity　of　263　true　external　compounds　without　experimental　values　against　Nsp14,　and　the　prediction　reliability　was　analyzed　and　discussed.　Molecular　docking　and　ADMET　analyses　were　conducted　for　compounds　with　higher　similarity　to　the　modelling　compounds.　Finally,　two　compounds　were　identified　as　potential　candidate　drugs　of　targeting　Nsp14.　The　current　work　lays　a　solid　theoretical　foundation　for　the　discovery　of　inhibitors　targeting　Nsp14,　and　has　an　important　reference　significance　for　the　development　of　anti-COVID-19　drugs.