diff --git a/datasette/views/table.py b/datasette/views/table.py
index 139ff80b..8c24bf53 100644
--- a/datasette/views/table.py
+++ b/datasette/views/table.py
@@ -63,6 +63,26 @@ class Row:
 
 
 class RowTableShared(DataView):
+    async def columns_to_select(self, db, table, request):
+        table_columns = await db.table_columns(table)
+        if "_col" in request.args and "_nocol" in request.args:
+            raise DatasetteError("Cannot use _col and _nocol at the same time")
+        if "_col" in request.args:
+            new_columns = []
+            for column in request.args["_col"]:
+                if column not in table_columns:
+                    raise DatasetteError("_col={} is an invalid column".format(column))
+                new_columns.append(column)
+            return new_columns
+        elif "_nocol" in request.args:
+            # Return all columns EXCEPT these
+            bad_columns = [column for column in request.args["_nocol"] if column not in table_columns]
+            if bad_columns:
+                raise DatasetteError("_nocol={} - invalid columns".format(", ".join(bad_columns)))
+            return [column for column in table_columns if column not in request.args["_nocol"]]
+        else:
+            return table_columns
+
     async def sortable_columns_for_table(self, database, table, use_rowid):
         db = self.ds.databases[database]
         table_metadata = self.ds.table_metadata(database, table)
@@ -235,17 +255,18 @@ class TableView(RowTableShared):
             raise NotFound("Table not found: {}".format(table))
 
         pks = await db.primary_keys(table)
-        table_columns = await db.table_columns(table)
 
-        select_columns = ", ".join(escape_sqlite(t) for t in table_columns)
+        # Take _col= and _nocol= into account
+        table_columns = await self.columns_to_select(db, table, request)
+        select_clause = ", ".join(escape_sqlite(t) for t in table_columns)
 
         use_rowid = not pks and not is_view
         if use_rowid:
-            select = "rowid, {}".format(select_columns)
+            select = "rowid, {}".format(select_clause)
             order_by = "rowid"
             order_by_pks = "rowid"
         else:
-            select = select_columns
+            select = select_clause
             order_by_pks = ", ".join([escape_sqlite(pk) for pk in pks])
             order_by = order_by_pks
 
@@ -261,12 +282,10 @@ class TableView(RowTableShared):
         # That's so if there is a column that starts with _
         # it can still be queried using ?_col__exact=blah
         special_args = {}
-        special_args_lists = {}
         other_args = []
         for key, value in args.items():
             if key.startswith("_") and "__" not in key:
                 special_args[key] = value[0]
-                special_args_lists[key] = value
             else:
                 for v in value:
                     other_args.append((key, v))
@@ -499,18 +518,6 @@ class TableView(RowTableShared):
         if order_by:
             order_by = "order by {} ".format(order_by)
 
-        # _group_count=col1&_group_count=col2
-        group_count = special_args_lists.get("_group_count") or []
-        if group_count:
-            sql = 'select {group_cols}, count(*) as "count" from {table_name} {where} group by {group_cols} order by "count" desc limit 100'.format(
-                group_cols=", ".join(
-                    '"{}"'.format(group_count_col) for group_count_col in group_count
-                ),
-                table_name=escape_sqlite(table),
-                where=where_clause,
-            )
-            return await self.custom_sql(request, database, hash, sql, editable=True)
-
         extra_args = {}
         # Handle ?_size=500
         page_size = _size or request.raw_args.get("_size")
@@ -598,7 +605,7 @@ class TableView(RowTableShared):
             facets_timed_out.extend(instance_facets_timed_out)
 
         # Figure out columns and rows for the query
-        columns = [r[0] for r in results.description]
+        columns = table_columns
         rows = list(results.rows)
 
         filter_columns = columns[:]
@@ -626,6 +633,8 @@ class TableView(RowTableShared):
                 column = fk["column"]
                 if column not in columns_to_expand:
                     continue
+                if column not in columns:
+                    continue
                 expanded_columns.append(column)
                 # Gather the values
                 column_index = columns.index(column)
diff --git a/docs/json_api.rst b/docs/json_api.rst
index de70362c..e369bee7 100644
--- a/docs/json_api.rst
+++ b/docs/json_api.rst
@@ -321,15 +321,6 @@ Special table arguments
 
     Here's `an example <https://latest.datasette.io/fixtures/roadside_attractions?_through={%22table%22:%22roadside_attraction_characteristics%22,%22column%22:%22characteristic_id%22,%22value%22:%221%22}>`__.
 
-
-``?_group_count=COLUMN``
-    Executes a SQL query that returns a count of the number of rows matching
-    each unique value in that column, with the most common ordered first.
-
-``?_group_count=COLUMN1&_group_count=column2``
-    You can pass multiple ``_group_count`` columns to return counts against
-    unique combinations of those columns.
-
 ``?_next=TOKEN``
     Pagination by continuation token - pass the token that was returned in the
     ``"next"`` property by the previous page.