diff --git a/dataset/persistence/database.py b/dataset/persistence/database.py
index b03239f..a46e0a4 100644
--- a/dataset/persistence/database.py
+++ b/dataset/persistence/database.py
@@ -79,6 +79,11 @@ class Database(object):
         """Return a SQLAlchemy schema cache object."""
         return MetaData(schema=self.schema, bind=self.executable)
 
+    @property
+    def in_transaction(self):
+        """Check if this database is in a transactional context."""
+        return len(self.local.tx) > 0
+
     def _flush_tables(self):
         """Clear the table metadata after transaction rollbacks."""
         for table in self._tables.values():
diff --git a/dataset/persistence/table.py b/dataset/persistence/table.py
index 2d3291e..98cf4f7 100644
--- a/dataset/persistence/table.py
+++ b/dataset/persistence/table.py
@@ -1,4 +1,6 @@
 import logging
+import warnings
+import threading
 
 from sqlalchemy.sql import and_, expression
 from sqlalchemy.sql.expression import ClauseElement
@@ -207,6 +209,13 @@ class Table(object):
             except NoSuchTableError:
                 pass
 
+    def _threading_warn(self):
+        if self.db.in_transaction and threading.active_count() > 1:
+            warnings.warn("Changing the database schema inside a transaction "
+                          "in a multi-threaded environment is likely to lead "
+                          "to race conditions and synchronization issues.",
+                          RuntimeWarning)
+
     def _sync_table(self, columns):
         """Lazy load, create or adapt the table structure in the database."""
         if self._table is None:
@@ -218,6 +227,7 @@ class Table(object):
                 raise DatasetException("Table does not exist: %s" % self.name)
             # Keep the lock scope small because this is run very often.
             with self.db.lock:
+                self._threading_warn()
                 self._table = SQLATable(self.name,
                                         self.db.metadata,
                                         schema=self.db.schema)
@@ -236,6 +246,7 @@ class Table(object):
                 self._table.create(self.db.executable, checkfirst=True)
         elif len(columns):
             with self.db.lock:
+                self._threading_warn()
                 for column in columns:
                     self.db.op.add_column(self.name, column, self.db.schema)
                 self._reflect_table()
@@ -317,7 +328,7 @@ class Table(object):
                 log.debug("Column exists: %s" % name)
                 return
 
-            log.debug("Create column: %s on %s", name, self.name)
+            self._threading_warn()
             self.db.op.add_column(
                 self.name,
                 Column(name, type),
@@ -353,6 +364,7 @@ class Table(object):
                 log.debug("Column does not exist: %s", name)
                 return
 
+            self._threading_warn()
             self.db.op.drop_column(
                 self.table.name,
                 name,
@@ -367,6 +379,7 @@ class Table(object):
         """
         with self.db.lock:
             if self.exists:
+                self._threading_warn()
                 self.table.drop(self.db.executable, checkfirst=True)
                 self._table = None
 
@@ -401,6 +414,7 @@ class Table(object):
                 raise DatasetException("Table has not been created yet.")
 
             if not self.has_index(columns):
+                self._threading_warn()
                 name = name or index_name(self.name, columns)
                 columns = [self.table.c[c] for c in columns]
                 idx = Index(name, *columns, **kw)