diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7218b95..25680d3 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,6 +10,7 @@ folgen [Semantic Versioning](https://semver.org/lang/de/). Die Version steht in
 ## [0.6.0] - 2026-10-11
 
 ### Hinzugefügt
+- `column -d Spalte ...` (und `project -d`) löscht Spalten: alle anderen bleiben in ihrer Reihenfolge. Fehler bei unbekannter Spalte oder wenn alle gelöscht würden. Dazu Abschnitt in `man/column.man`, Beispiel in `testall.cmds` und Tests.
 - Steuertabellen 2026 für `tax`: `demo/taxDE2026`, `taxAT2026`, `taxCH2026` (nur Bundessteuer), `taxLI2026` (nur Landessteuer), erzeugt mit `tools/gen_tax_tables.rb`, mit Tests.
 
 ### Geändert
diff --git a/README.md b/README.md
index 7561e2a..180b910 100644
--- a/README.md
+++ b/README.md
@@ -40,6 +40,8 @@ Dasselbe gilt für **Listen** (ein Datensatz als `Spalte<Tab>Wert`-Zeilen, die
 Datei beginnt mit einer Leerzeile). `tabletolist` und `listtotable` wandeln um,
 die meisten Befehle lesen beides.
 
+`column -d Spalte ...` löscht Spalten (alle anderen bleiben), `column Spalte ...` wählt sie aus.
+
 `row`, `compute` und `validate` übergeben ihren Ausdruck an awk, nachdem sie
 Spaltennamen durch `$N` ersetzt haben. Es gilt also awk-Syntax.
 
diff --git a/demo/testall.cmds b/demo/testall.cmds
index 5d653e4..fad1f19 100644
--- a/demo/testall.cmds
+++ b/demo/testall.cmds
@@ -12,6 +12,9 @@ rdb select 'Description ~ /gloves/' < inventory
 ## column (project): pick columns in any order; a name that does not exist becomes a new empty column
 column Description Cost < inventory
 project Item Notes < owners
+## column -d: keep all columns except the named ones (a deletion)
+column -d Cost Value < inventory
+subtotal Date Debit Credit < journal | column -d Account Description
 ## sorttable: -n numeric, -r reverse, -f fold case
 sorttable -nr Cost < inventory
 ## compute: calculate columns from other columns (the new column must exist, so make it with addcol)
diff --git a/demo/testall.old b/demo/testall.old
index 84296f3..c032c61 100644
--- a/demo/testall.old
+++ b/demo/testall.old
@@ -28,6 +28,23 @@ Item	Notes
 1	
 3	
 6	
+## column -d: keep all columns except the named ones (a deletion)
+$ column -d Cost Value < inventory
+Item	Amount	Description
+----	------	-----------
+1	3	rubber gloves
+2	100	test tubes
+3	5	clamps
+4	23	plates
+5	99	cleaning cloth
+6	89	bunsen burners
+7	5	scales
+$ subtotal Date Debit Credit < journal | column -d Account Description
+Date	Debit	Credit
+----	-----	------
+861222	25000	25000
+861223	10000	10000
+861224	30000	30000
 ## sorttable: -n numeric, -r reverse, -f fold case
 $ sorttable -nr Cost < inventory
 Item	Amount	Cost	Value	Description
diff --git a/lib/rdb/cmd/relational.rb b/lib/rdb/cmd/relational.rb
index 1e32988..28171ac 100644
--- a/lib/rdb/cmd/relational.rb
+++ b/lib/rdb/cmd/relational.rb
@@ -11,9 +11,20 @@ Rdb.register('row', 'Make a new table where rows match logical condition',
 end
 Rdb.alias_for('select', 'row')
 
-Rdb.register('column', 'Display columns of a table in any order', 'column [ Column ... ] < tableorlist') do |args|
+Rdb.register('column', 'Display columns of a table in any order', 'column [ -d ] [ Column ... ] < tableorlist') do |args|
+  drop = args.first == '-d'
+  args.shift if drop
   t = table_in
-  next t.write if args.empty?
+  next t.write if args.empty? && !drop
+  die 'usage: column -d Column ... < tableorlist' if drop && args.empty?
+
+  if drop # keep every column except the named ones
+    gone = args.map { |a| t.index_of(a) }
+    keep = t.head.each_index.reject { |i| gone.include?(i) }
+    die 'column -d would delete every column' if keep.empty?
+
+    next t.with(head: keep.map { |i| t.head[i] }, rows: t.rows.map { |r| keep.map { |i| r[i] } }).write
+  end
 
   idx = args.map { |a| Rdb.column_index(t.head, a) }
   head = args.each_with_index.map { |a, n| idx[n] ? t.head[idx[n]] : a }
diff --git a/man/column.man b/man/column.man
index 2cac9e6..003baa9 100644
--- a/man/column.man
+++ b/man/column.man
@@ -129,6 +129,24 @@ Names that are also
 reserved words or are followed by
 .B (
 are never taken for columns.
+.SH "Deleting Columns"
+.I "(Ruby re-creation only.)"
+.DM
+column -d Column ... < tableorlist
+.DE
+.PP
+With the option
+.BR -d ,
+.B column
+keeps every column
+.I except
+those named.
+The order of the other columns stays as it is.
+It is an error to name a column that does not exist,
+or to delete every column.
+.DM
+subtotal Date Debit Credit < journal | column -d Account Description
+.DE
 .SH "See Also"
 .B
 project,
diff --git a/test/rdb_test.rb b/test/rdb_test.rb
index 4ef46f3..cc665e3 100644
--- a/test/rdb_test.rb
+++ b/test/rdb_test.rb
@@ -5,6 +5,31 @@ require 'pty'
 require_relative '../lib/rdb'
 
 class SelectionTest < RdbCase
+  def test_column_d_deletes_columns
+    out = rdb('column', '-d', 'Cost', 'Value', file: 'inventory')
+    assert_equal %w[Item Amount Description], out.lines.first.chomp.split("\t")
+    assert_equal %w[1 3 rubber\ gloves], data(out).first
+    assert_equal %w[Date Debit Credit], rdb('column', '-d', 'Account', 'Description', stdin: rdb('subtotal', 'Date', 'Debit', 'Credit', file: 'journal')).lines.first.chomp.split("\t")
+  end
+
+  def test_column_d_errors
+    src = File.read(File.join(@dir, 'owners'))
+    [%w[-d], %w[-d Nope], %w[-d Item Owner]].each do |args|
+      out, err, st = run3('column', *args, stdin: src)
+      refute st.success?, "column #{args.join(' ')} should fail"
+      assert_equal '', out
+      refute_empty err
+    end
+  end
+
+  def test_column_d_works_on_lists_and_with_icase
+    list = rdb('tabletolist', file: 'owners')
+    assert_includes rdb('column', '-d', 'Item', stdin: list), "Owner\tann"
+    out, _e, st = run3('column', '-d', 'COST', stdin: File.read(File.join(@dir, 'inventory')), env: { 'RDB_ICASE' => '1' })
+    assert st.success?
+    refute_includes out.lines.first, 'Cost'
+  end
+
   def test_splittable_pieces_and_errors
     out, = run3('splittable', '-4', 'inventory')
     assert_equal 4, data(File.read(File.join(@dir, 'inventoryaa'))).size
@@ -463,7 +488,7 @@ end
 
 class HelpTest < RdbCase
   def test_whatis_whatwill_commands
-    assert_includes rdb('whatis', 'column'), "   column    - display columns of a table in any order\n   column    column [ Column ... ] < tableorlist\n"
+    assert_includes rdb('whatis', 'column'), "   column    - display columns of a table in any order\n   column    column [ -d ] [ Column ... ] < tableorlist\n"
     assert_includes rdb('whatwill', 'total'), 'subtotal'
     assert_operator data(rdb('commands')).size, :>, 120
     assert_operator rdb('howmany').to_i, :>, 120