diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8e75b23..84797b7 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -4,6 +4,7 @@ This document contains the Spec2nii release history in reverse chronological ord
 --------------------------------
 - The --anon flag can be passed with any call to anonymise after writing files.
 - The Siemens enhanced dicom filetype pathway now handles CSI data.
+- Fixed issue with RDA files having latin1 encoding. Thanks to gaunab on github. Fixes Issue #96.
 
 0.7.0 (Saturday 5th August 2023)
 --------------------------------
diff --git a/spec2nii/Siemens/rda.py b/spec2nii/Siemens/rda.py
index a3cc25e..294948f 100644
--- a/spec2nii/Siemens/rda.py
+++ b/spec2nii/Siemens/rda.py
@@ -44,18 +44,33 @@ def convert_rda(rda_path, fname_out, verbose):
 
     with open(rda_path, 'rb') as fp:
         for line in fp:
-            if hdr_st.search(line.decode()):
-                pass
-                # print('header found')
-            elif hdr_end.search(line.decode()):
-                # print('header end')
-                break
-            else:
-                match = hdr_val.search(line.decode())
-                if len(match.groups()) < 2:
-                    hdr[match[1]] = None
+            try:
+                if hdr_st.search(line.decode()):
+                    pass
+                    # print('header found')
+                elif hdr_end.search(line.decode()):
+                    # print('header end')
+                    break
+                else:
+                    match = hdr_val.search(line.decode())
+                    if len(match.groups()) < 2:
+                        hdr[match[1]] = None
+                    else:
+                        hdr[match[1]] = match[2]
+            except UnicodeDecodeError:
+                print('Trying latin-1 encoding.')
+                if hdr_st.search(line.decode('latin-1')):
+                    pass
+                    # print('header found')
+                elif hdr_end.search(line.decode('latin-1')):
+                    # print('header end')
+                    break
                 else:
-                    hdr[match[1]] = match[2]
+                    match = hdr_val.search(line.decode('latin-1'))
+                    if len(match.groups()) < 2:
+                        hdr[match[1]] = None
+                    else:
+                        hdr[match[1]] = match[2]
         if verbose:
             print(hdr)
 
diff --git a/tests/spec2nii_test_data b/tests/spec2nii_test_data
index 6f205d3..e0ad587 160000
--- a/tests/spec2nii_test_data
+++ b/tests/spec2nii_test_data
@@ -1 +1 @@
-Subproject commit 6f205d389085a420b38689f573045166517bb445
+Subproject commit e0ad5870cd5891bbe3fdcf9415845cde1f194884
diff --git a/tests/test_siemens_rda.py b/tests/test_siemens_rda.py
index aac48e4..b8f2be8 100644
--- a/tests/test_siemens_rda.py
+++ b/tests/test_siemens_rda.py
@@ -18,6 +18,8 @@
 xa20_svs_path = siemens_path / 'XAData' / 'XA20/rda/spct_002.MR.MRI-LAB Test_Dir.5.1.114540.rda'
 xa31_locale_svs_path = siemens_path / 'XAData' / 'XA31/rda/locale_XA31.rda'
 
+latin1_encoding = siemens_path / 'rda' / 'latin1.rda'
+
 
 def test_xa20_svs(tmp_path):
 
@@ -53,3 +55,21 @@ def test_xa31_locale_svs(tmp_path):
 
     assert img_t.shape == (1, 1, 1, 1024)
     assert np.iscomplexobj(img_t.dataobj)
+
+
+def test_latin_encoding(tmp_path):
+
+    subprocess.check_call(['spec2nii', 'rda',
+                           '-f', 'rda_latin1',
+                           '-o', tmp_path,
+                           '-j', str(latin1_encoding)])
+
+    img_t = read_nifti_mrs(tmp_path / 'rda_latin1.nii.gz')
+
+    hdr_ext_codes = img_t.header.extensions.get_codes()
+    hdr_ext = json.loads(img_t.header.extensions[hdr_ext_codes.index(44)].get_content())
+
+    hdr_ext['ResonantNucleus'] = ['1H', ]
+
+    assert img_t.shape == (1, 1, 1, 2048)
+    assert np.iscomplexobj(img_t.dataobj)