diff --git a/icu4c/source/common/rbbidata.cpp b/icu4c/source/common/rbbidata.cpp
index 09d47964e61..271553a89c1 100644
--- a/icu4c/source/common/rbbidata.cpp
+++ b/icu4c/source/common/rbbidata.cpp
@@ -1,6 +1,6 @@
 /*
 ***************************************************************************
-*   Copyright (C) 1999-2010 International Business Machines Corporation   *
+*   Copyright (C) 1999-2014 International Business Machines Corporation   *
 *   and others. All rights reserved.                                      *
 ***************************************************************************
 */
@@ -46,20 +46,39 @@ U_NAMESPACE_BEGIN
 //
 //-----------------------------------------------------------------------------
 RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status) {
+    init0();
     init(data, status);
 }
 
 RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt, UErrorCode &status) {
+    init0();
     init(data, status);
     fDontFreeData = TRUE;
 }
 
 RBBIDataWrapper::RBBIDataWrapper(UDataMemory* udm, UErrorCode &status) {
-    const RBBIDataHeader *d = (const RBBIDataHeader *)
-        // ((char *)&(udm->pHeader->info) + udm->pHeader->info.size);
-        // taking into consideration the padding added in by udata_write
-        ((char *)(udm->pHeader) + udm->pHeader->dataHeader.headerSize);
-    init(d, status);
+    init0();
+    if (U_FAILURE(status)) {
+        return;
+    }
+    const DataHeader *dh = udm->pHeader;
+    int32_t headerSize = dh->dataHeader.headerSize;
+    if (  !(headerSize >= 20 &&
+            dh->info.isBigEndian == U_IS_BIG_ENDIAN &&
+            dh->info.charsetFamily == U_CHARSET_FAMILY &&
+            dh->info.dataFormat[0] == 0x42 &&  // dataFormat="Brk "
+            dh->info.dataFormat[1] == 0x72 &&
+            dh->info.dataFormat[2] == 0x6b &&
+            dh->info.dataFormat[3] == 0x20)
+            // Note: info.fFormatVersion is duplicated in the RBBIDataHeader, and is
+            //       validated when checking that.
+        ) {
+        status = U_INVALID_FORMAT_ERROR;
+        return;
+    }
+    const char *dataAsBytes = reinterpret_cast<const char *>(dh);
+    const RBBIDataHeader *rbbidh = reinterpret_cast<const RBBIDataHeader *>(dataAsBytes + headerSize);
+    init(rbbidh, status);
     fUDataMem = udm;
 }
 
@@ -69,6 +88,19 @@ RBBIDataWrapper::RBBIDataWrapper(UDataMemory* udm, UErrorCode &status) {
 //              constructors.
 //
 //-----------------------------------------------------------------------------
+void RBBIDataWrapper::init0() {
+    fHeader = NULL;
+    fForwardTable = NULL;
+    fReverseTable = NULL;
+    fSafeFwdTable = NULL;
+    fSafeRevTable = NULL;
+    fRuleSource = NULL;
+    fRuleStatusTable = NULL;
+    fUDataMem = NULL;
+    fRefCount = 0;
+    fDontFreeData = TRUE;
+}
+
 void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
     if (U_FAILURE(status)) {
         return;
@@ -84,10 +116,6 @@ void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
     //       an int32_t field, rather than an array of 4 bytes.
 
     fDontFreeData = FALSE;
-    fUDataMem     = NULL;
-    fReverseTable = NULL;
-    fSafeFwdTable = NULL;
-    fSafeRevTable = NULL;
     if (data->fFTableLen != 0) {
         fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable);
     }
diff --git a/icu4c/source/common/rbbidata.h b/icu4c/source/common/rbbidata.h
index 78a4ac762c8..65274c3adbb 100644
--- a/icu4c/source/common/rbbidata.h
+++ b/icu4c/source/common/rbbidata.h
@@ -1,7 +1,7 @@
 /*
 *******************************************************************************
 *
-*   Copyright (C) 1999-2013 International Business Machines
+*   Copyright (C) 1999-2014 International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
@@ -150,6 +150,7 @@ public:
     RBBIDataWrapper(UDataMemory* udm, UErrorCode &status);
     ~RBBIDataWrapper();
 
+    void                  init0();
     void                  init(const RBBIDataHeader *data, UErrorCode &status);
     RBBIDataWrapper      *addReference();
     void                  removeReference();
@@ -182,7 +183,7 @@ public:
 
 private:
     u_atomic_int32_t    fRefCount;
-    UDataMemory        *fUDataMem;
+    UDataMemory  *fUDataMem;
     UnicodeString       fRuleString;
     UBool               fDontFreeData;
 
diff --git a/icu4c/source/common/unicode/udata.h b/icu4c/source/common/unicode/udata.h
index a95fe16bcb6..29e46630d53 100644
--- a/icu4c/source/common/unicode/udata.h
+++ b/icu4c/source/common/unicode/udata.h
@@ -96,7 +96,7 @@ U_CDECL_BEGIN
  *
  * <p>The <code>formatVersion</code> field should be used to
  * make sure that the format can be interpreted.
- * I may be a good idea to check only for the one or two highest
+ * It may be a good idea to check only for the one or two highest
  * of the version elements to allow the data memory to
  * get more or somewhat rearranged contents, for as long
  * as the using code can still interpret the older contents.</p>