In　this　paper,　the　fiber　laser　was　used　to　cut　stainless　steel　sheet　of　thickness　of　0.01　mm　into　8　x　8　mm　mask-plates,　and　the　acceptable　tolerances　is　no　more　than　+/-　10　mu　m.　The　cutting　quality　and　related　affecting　parameters　were　analyzed,　such　as　kerf　width,　cut　surface　quality,　heat-affected-zone　(HAZ),　laser　power,　cutting　speed　and　assist　gas　and　it＇s　pressure.　In　order　to　obtain　the　excellent　result　of　the　laser　micro-cutting,　the　radial　basis　function　(RBF)　neural　network　was　analyzed　and　employed　to　build　a　predicting　model.　With　the　trained　RBF　neural　network,　the　micro-cutting　process　parameters　were　predicted　and　the　micro-cutting　process　was　simulated,　the　parameters　obtaining　good　simulate　result　were　used　in　actual　micro-cutting　process.　A　good　cutting　quality　mask-plate　was　obtained　by　using　predicting　parameters　of　RBF　neural　network　model　in　the　experiment.　The　results　of　the　experiment　reveal　that　the　RBF　neural　network　is　an　intelligent　method　to　optimize　parameters　for　laser　micro-cutting.