diff --git a/moo.js b/moo.js
index 7418b74..0e9e22d 100644
--- a/moo.js
+++ b/moo.js
@@ -41,8 +41,6 @@
       return '(?:' + reEscape(obj) + ')'
 
     } else if (isRegExp(obj)) {
-      // TODO: consider /u support
-      if (obj.ignoreCase) throw new Error('RegExp /i flag not allowed')
       if (obj.global) throw new Error('RegExp /g flag is implied')
       if (obj.sticky) throw new Error('RegExp /y flag is implied')
       if (obj.multiline) throw new Error('RegExp /m flag is implied')
@@ -120,6 +118,7 @@
       value: null,
       type: null,
       shouldThrow: false,
+      ignoreCase: null,
     }
 
     // Avoid Object.assign(), so we support IE9+
@@ -154,6 +153,7 @@
     var fast = Object.create(null)
     var fastAllowed = true
     var unicodeFlag = null
+    var ignoreCaseFlag = null
     var groups = []
     var parts = []
 
@@ -210,9 +210,13 @@
 
       groups.push(options)
 
-      // Check unicode flag is used everywhere or nowhere
+      // Check unicode and ignoreCase flags are used everywhere or nowhere
+      var hasLiteralsWithCase = false
       for (var j = 0; j < match.length; j++) {
         var obj = match[j]
+        if (typeof obj === "string" && obj.toLowerCase() !== obj.toUpperCase()) {
+          hasLiteralsWithCase = true
+        }
         if (!isRegExp(obj)) {
           continue
         }
@@ -222,6 +226,31 @@
         } else if (unicodeFlag !== obj.unicode) {
           throw new Error("If one rule is /u then all must be")
         }
+
+        if (ignoreCaseFlag === null) {
+          ignoreCaseFlag = obj.ignoreCase
+        } else if (ignoreCaseFlag !== obj.ignoreCase) {
+          throw new Error("If one rule is /i then all must be")
+        }
+
+        // RegExp flags must match the rule's ignoreCase option, if set
+        if (options.ignoreCase !== null && obj.ignoreCase !== options.ignoreCase) {
+          throw new Error("ignoreCase option must match RegExp flags (in token '" + options.defaultType + "')")
+        }
+      }
+
+      if (hasLiteralsWithCase) {
+        var ignoreCase = !!options.ignoreCase
+        if (ignoreCaseFlag === null) {
+          ignoreCaseFlag = ignoreCase
+        } else if (ignoreCaseFlag !== ignoreCase) {
+          if (ignoreCaseFlag) {
+            throw new Error("Literal must be marked with {ignoreCase: true} (in token '" + options.defaultType + "')")
+          } else {
+            // TODO transform literals to ignore case, even if it's not set globally
+            throw new Error("If one rule sets ignoreCase then all must (in token '" + options.defaultType + "')")
+          }
+        }
       }
 
       // convert to RegExp
@@ -257,6 +286,7 @@
     var suffix = hasSticky || fallbackRule ? '' : '|'
 
     if (unicodeFlag === true) flags += "u"
+    if (ignoreCaseFlag === true) flags += "i"
     var combined = new RegExp(reUnion(parts) + suffix, flags)
     return {regexp: combined, groups: groups, fast: fast, error: errorRule || defaultErrorRule}
   }
diff --git a/test/test.js b/test/test.js
index 40158f6..d278616 100644
--- a/test/test.js
+++ b/test/test.js
@@ -28,10 +28,9 @@ describe('compiler', () => {
     expect(lex4.next()).toMatchObject({type: 'err', text: 'nope!'})
   })
 
-  test("warns for /g, /y, /i, /m", () => {
+  test("warns for /g, /y, /m", () => {
     expect(() => compile({ word: /foo/ })).not.toThrow()
     expect(() => compile({ word: /foo/g })).toThrow('implied')
-    expect(() => compile({ word: /foo/i })).toThrow('not allowed')
     expect(() => compile({ word: /foo/y })).toThrow('implied')
     expect(() => compile({ word: /foo/m })).toThrow('implied')
   })
@@ -1211,3 +1210,84 @@ describe("unicode flag", () => {
   })
 
 })
+
+
+describe('ignoreCase flag', () => {
+
+  test("allows all rules to be /i", () => {
+    expect(() => compile({ a: /foo/i, b: /bar/i })).not.toThrow()
+    expect(() => compile({ a: /foo/i, b: /bar/ })).toThrow("If one rule is /i then all must be")
+    expect(() => compile({ a: /foo/, b: /bar/i })).toThrow("If one rule is /i then all must be")
+  })
+
+  test("allows all rules to be /ui", () => {
+    expect(() => compile({ a: /foo/ui, b: /bar/ui })).not.toThrow()
+    expect(() => compile({ a: /foo/u, b: /bar/i })).toThrow("If one rule is /u then all must be")
+    expect(() => compile({ a: /foo/i, b: /bar/u })).toThrow("If one rule is /u then all must be")
+    expect(() => compile({ a: /foo/ui, b: /bar/i })).toThrow("If one rule is /u then all must be")
+    expect(() => compile({ a: /foo/ui, b: /bar/u })).toThrow("If one rule is /i then all must be")
+    expect(() => compile({ a: /foo/i, b: /bar/ui })).toThrow("If one rule is /u then all must be")
+    expect(() => compile({ a: /foo/u, b: /bar/ui })).toThrow("If one rule is /i then all must be")
+  })
+
+  test("allow literals to be marked ignoreCase", () => {
+    expect(() => compile({
+      a: /foo/i,
+      lit: {match: "quxx", ignoreCase: true},
+    })).not.toThrow()
+    expect(() => compile([
+      { type: "a", match: /foo/i },
+      { type: "lit", match: "quxx", ignoreCase: true },
+    ])).not.toThrow()
+  })
+
+  test("require literals to be marked ignoreCase", () => {
+    expect(() => compile({
+      a: /foo/i,
+      lit: "quxx" ,
+    })).toThrow("Literal must be marked with {ignoreCase: true} (in token 'lit')")
+    expect(() => compile([
+      { type: "a", match: /foo/i },
+      { type: "lit", match: "quxx" },
+    ])).toThrow("Literal must be marked with {ignoreCase: true} (in token 'lit')")
+  })
+
+  test("ignoreCase is only required when case is relevant", () => {
+    expect(() => compile({
+      cat: {match: "cat", ignoreCase: true},
+      bat: {match: "BAT", ignoreCase: true},
+      comma: ',',
+      semi: ';',
+      lparen: '(',
+      rparen: ')',
+      lbrace: '{',
+      rbrace: '}',
+      lbracket: '[',
+      rbracket: ']',
+      and: '&&',
+      or: '||',
+      bitand: '&',
+      bitor: '|',
+    })).not.toThrow()
+  })
+
+  test("require ignoreCase option to be match RegExp flags", () => {
+    expect(() => compile({
+      word: { match: /[a-z]+/, ignoreCase: true },
+    })).toThrow("ignoreCase option must match RegExp flags")
+    expect(() => compile({
+      word: { match: ["foo", /[a-z]+/], ignoreCase: true },
+    })).toThrow("ignoreCase option must match RegExp flags")
+    expect(() => compile({
+      word: { match: /[a-z]+/i, ignoreCase: false },
+    })).toThrow("ignoreCase option must match RegExp flags")
+  })
+
+  test("supports ignoreCase", () => {
+    const lexer = compile({ a: /foo/i, b: /bar/i, })
+    lexer.reset("FoObAr")
+    expect(lexer.next()).toMatchObject({value: "FoO"})
+    expect(lexer.next()).toMatchObject({value: "bAr"})
+  })
+
+})