diff --git a/tests/invalid_unicode_sqlite.vd b/tests/invalid_unicode_sqlite.vd
index 47f959bdf..0a11b805d 100644
--- a/tests/invalid_unicode_sqlite.vd
+++ b/tests/invalid_unicode_sqlite.vd
@@ -1,5 +1,7 @@
 sheet	col	row	longname	input	keystrokes	comment
 global		encoding	set-option	latin-1		
 invalid_unicode		encoding	set-option	latin-1		
+global		save_encoding	set-option	latin-1		
+invalid_unicode		save_encoding	set-option	latin-1		
 			open-file	tests/invalid_unicode.sqlite	o	
 invalid_unicode		キTest	open-row		Enter	open sheet with copies of rows referenced in current row
diff --git a/visidata/cmdlog.py b/visidata/cmdlog.py
index 55289c29c..b0a7a354d 100644
--- a/visidata/cmdlog.py
+++ b/visidata/cmdlog.py
@@ -34,7 +34,7 @@ def open_vdj(vd, p):
 
 @VisiData.api
 def save_vdj(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         fp.write("#!vd -p\n")
         for vs in vsheets:
             vs.write_jsonl(fp)
diff --git a/visidata/loaders/csv.py b/visidata/loaders/csv.py
index b370fcb63..b3d66ad9e 100644
--- a/visidata/loaders/csv.py
+++ b/visidata/loaders/csv.py
@@ -58,7 +58,7 @@ def iterload(self):
 @VisiData.api
 def save_csv(vd, p, sheet):
     'Save as single CSV file, handling column names as first line.'
-    with p.open_text(mode='w', encoding=sheet.options.encoding, newline='') as fp:
+    with p.open_text(mode='w', encoding=sheet.options.save_encoding, newline='') as fp:
         cw = csv.writer(fp, **options.getall('csv_'))
         colnames = [col.name for col in sheet.visibleCols]
         if ''.join(colnames):
diff --git a/visidata/loaders/fixed_width.py b/visidata/loaders/fixed_width.py
index 359a12938..04d81ceef 100644
--- a/visidata/loaders/fixed_width.py
+++ b/visidata/loaders/fixed_width.py
@@ -76,7 +76,7 @@ def setCols(self, headerlines):
 
 @VisiData.api
 def save_fixed(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         for sheet in vsheets:
             if len(vsheets) > 1:
                 fp.write('%s\n\n' % sheet.name)
diff --git a/visidata/loaders/jrnl.py b/visidata/loaders/jrnl.py
index 688006963..40555c2b4 100644
--- a/visidata/loaders/jrnl.py
+++ b/visidata/loaders/jrnl.py
@@ -42,7 +42,7 @@ def iterload(self):
 
 @VisiData.api
 def save_jrnl(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         for vs in vsheets:
             for r in vs.iterrows():
                 fp.write(f'[{r.date} {r.time}] {r.title}\n')
diff --git a/visidata/loaders/json.py b/visidata/loaders/json.py
index bda34eb51..c86587f23 100644
--- a/visidata/loaders/json.py
+++ b/visidata/loaders/json.py
@@ -106,7 +106,7 @@ def encode_json(vd, row, cols, enc=_vjsonEncoder(sort_keys=False)):
 @VisiData.api
 def save_json(vd, p, *vsheets):
     vs = vsheets[0]
-    with p.open_text(mode='w', encoding=vs.options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vs.options.save_encoding) as fp:
         try:
             indent = int(vs.options.json_indent)
         except Exception:
@@ -151,7 +151,7 @@ def write_jsonl(vs, fp):
 
 @VisiData.api
 def save_jsonl(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         for vs in vsheets:
             vs.write_jsonl(fp)
 
diff --git a/visidata/loaders/jsonla.py b/visidata/loaders/jsonla.py
index 321137d7c..b75fcfaf8 100644
--- a/visidata/loaders/jsonla.py
+++ b/visidata/loaders/jsonla.py
@@ -66,6 +66,6 @@ def write_jsonla(vs, fp):
 
 @VisiData.api
 def save_jsonla(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         for vs in vsheets:
             write_jsonla(vs, fp)
diff --git a/visidata/loaders/markdown.py b/visidata/loaders/markdown.py
index d69296981..f050b8b2e 100644
--- a/visidata/loaders/markdown.py
+++ b/visidata/loaders/markdown.py
@@ -26,7 +26,7 @@ def write_md(p, *vsheets, md_style='orgmode'):
     else:
         delim = '|'
 
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         for vs in vsheets:
             if len(vsheets) > 1:
                 fp.write('# %s\n\n' % vs.name)
diff --git a/visidata/loaders/orgmode.py b/visidata/loaders/orgmode.py
index 4a4fb2add..05ebea383 100644
--- a/visidata/loaders/orgmode.py
+++ b/visidata/loaders/orgmode.py
@@ -359,7 +359,7 @@ def sysopen_row(sheet, row):
 
 @VisiData.api
 def save_org(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         for vs in vsheets:
             if isinstance(vs, OrgSheet):
                 for row in vs.rows:
diff --git a/visidata/loaders/tsv.py b/visidata/loaders/tsv.py
index d8096e020..d69e0d751 100644
--- a/visidata/loaders/tsv.py
+++ b/visidata/loaders/tsv.py
@@ -44,7 +44,7 @@ def iterload(self):
         delim = self.delimiter or self.options.delimiter
         rowdelim = self.row_delimiter or self.options.row_delimiter
 
-        with self.source.open_text(encoding=self.options.encoding) as fp:
+        with self.source.open_text(encoding=self.options.save_encoding) as fp:
                 for line in splitter(fp, rowdelim):
                     if not line:
                         continue
@@ -65,7 +65,7 @@ def save_tsv(vd, p, vs, delimiter='', row_delimiter=''):
     rowsep = row_delimiter or vs.options.row_delimiter
     trdict = vs.safe_trdict()
 
-    with p.open_text(mode='w', encoding=vs.options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vs.options.save_encoding) as fp:
         colhdr = unitsep.join(col.name.translate(trdict) for col in vs.visibleCols) + rowsep
         fp.write(colhdr)
 
diff --git a/visidata/loaders/vdx.py b/visidata/loaders/vdx.py
index b36594511..75bc51ccc 100644
--- a/visidata/loaders/vdx.py
+++ b/visidata/loaders/vdx.py
@@ -22,7 +22,7 @@ def iterload(self):
 
 @VisiData.api
 def save_vdx(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         fp.write(f"# {visidata.__version_info__}\n")
         for vs in vsheets:
             prevrow = None
diff --git a/visidata/path.py b/visidata/path.py
index b47f4c930..b24a4b3d6 100644
--- a/visidata/path.py
+++ b/visidata/path.py
@@ -9,7 +9,7 @@
 
 from visidata import *
 
-vd.option('encoding', 'utf-8', 'encoding passed to codecs.open', replay=True)
+vd.option('encoding', 'utf-8', 'encoding passed to codecs.open when opening a file', replay=True)
 vd.option('encoding_errors', 'surrogateescape', 'encoding_errors passed to codecs.open', replay=True)
 
 @lru_cache()
diff --git a/visidata/save.py b/visidata/save.py
index cb11d870b..d2801a23a 100644
--- a/visidata/save.py
+++ b/visidata/save.py
@@ -6,6 +6,7 @@
 
 vd.option('confirm_overwrite', True, 'whether to prompt for overwrite confirmation on save')
 vd.option('safe_error', '#ERR', 'error string to use while saving', replay=True)
+vd.option('save_encoding', 'utf-8', 'encoding passed to codecs.open when saving a file', replay=True)
 
 @Sheet.api
 def safe_trdict(vs):
@@ -170,7 +171,7 @@ def save_zip(vd, p, *vsheets):
 
 @VisiData.api
 def save_txt(vd, p, *vsheets):
-    with p.open_text(mode='w', encoding=vsheets[0].options.encoding) as fp:
+    with p.open_text(mode='w', encoding=vsheets[0].options.save_encoding) as fp:
         for vs in vsheets:
             unitsep = vs.options.delimiter
             rowsep = vs.options.row_delimiter