diff --git a/src/rez/cli/_complete_util.py b/src/rez/cli/_complete_util.py
index c33637232..9f2ba7d6b 100644
--- a/src/rez/cli/_complete_util.py
+++ b/src/rez/cli/_complete_util.py
@@ -2,7 +2,7 @@
 import os.path
 from fnmatch import fnmatch
 from rez.vendor.argcomplete import CompletionFinder, default_validator, \
-    sys_encoding, split_line, debug
+    sys_encoding, split_line, debug, USING_PYTHON2
 
 
 class RezCompletionFinder(CompletionFinder):
@@ -12,9 +12,12 @@ def __init__(self, parser, comp_line, comp_point):
         self.exclude = None
         self.validator = default_validator
         self.wordbreaks = " \t\"'@><=;|&(:"  # TODO: might need to be configurable/OS specific
-
-        comp_point = len(comp_line[:comp_point].decode(sys_encoding))
-        comp_line = comp_line.decode(sys_encoding)
+        
+        if USING_PYTHON2:
+            comp_line = comp_line.decode(sys_encoding)
+            comp_point = len(comp_line[:comp_point].decode(sys_encoding))
+        else:
+            comp_point = len(comp_line.encode(sys_encoding)[:comp_point].decode(sys_encoding))
 
         cword_prequote, cword_prefix, cword_suffix, comp_words, \
             first_colon_pos = split_line(comp_line, comp_point)
diff --git a/src/rez/cli/complete.py b/src/rez/cli/complete.py
index ed62354ee..90b369da3 100644
--- a/src/rez/cli/complete.py
+++ b/src/rez/cli/complete.py
@@ -96,6 +96,7 @@ def _pop_arg(l, p):
                                     comp_line=comp_line,
                                     comp_point=comp_point)
     words = completer.completions
+    words = [w.decode() if hasattr(w, 'decode') else w for w in words]
     print(' '.join(words))