diff --git a/.travis.yml b/.travis.yml
index a451fb1..98eb6bf 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -1,6 +1,7 @@
 language: ruby
 rvm:
   - 2.3.0
+  - 2.7
 addons:
   code_climate:
     repo_token: bcecbf1b229a2ddd666a2c3830f26a0113fd56ae1586d30d2d3fb1af837bf0e4
diff --git a/Gemfile b/Gemfile
index 50e80a4..16467ff 100644
--- a/Gemfile
+++ b/Gemfile
@@ -2,9 +2,3 @@ source 'https://rubygems.org'
 
 # Specify your gem's dependencies in csv-importer.gemspec
 gemspec
-
-gem 'rspec', '~> 3.3.0'
-gem 'guard-rspec'
-gem 'activemodel'
-gem 'simplecov', require: nil
-gem "codeclimate-test-reporter", require: nil
diff --git a/csv-importer.gemspec b/csv-importer.gemspec
index fd9139a..7644368 100644
--- a/csv-importer.gemspec
+++ b/csv-importer.gemspec
@@ -20,6 +20,10 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency "virtus"
 
-  spec.add_development_dependency "bundler", "~> 1.8"
-  spec.add_development_dependency "rake", "~> 10.0"
+  spec.add_development_dependency "rspec", "~> 3.3.0"
+  spec.add_development_dependency "rake"
+  spec.add_development_dependency "guard-rspec"
+  spec.add_development_dependency "activemodel", "~> 5"
+  spec.add_development_dependency "simplecov"
+  spec.add_development_dependency "codeclimate-test-reporter"
 end
diff --git a/lib/csv_importer/csv_reader.rb b/lib/csv_importer/csv_reader.rb
index 22f690d..c5ece38 100644
--- a/lib/csv_importer/csv_reader.rb
+++ b/lib/csv_importer/csv_reader.rb
@@ -14,8 +14,12 @@ def csv_rows
       @csv_rows ||= begin
         sane_content = sanitize_content(read_content)
         separator = detect_separator(sane_content)
-        cells = CSV.parse(sane_content, col_sep: separator, quote_char: quote_char, skip_blanks: true, encoding: encoding)
-        sanitize_cells(cells)
+        cells = CSV.parse(
+          sane_content,
+          col_sep: separator, quote_char: quote_char, skip_blanks: true,
+          external_encoding: source_encoding
+        )
+        sanitize_cells(encode_cells(cells))
       end
     end
 
@@ -44,9 +48,8 @@ def read_content
     end
 
     def sanitize_content(csv_content)
-      internal_encoding = encoding.split(':').last
       csv_content
-        .encode(Encoding.find(internal_encoding), {invalid: :replace, undef: :replace, replace: ''}) # Remove invalid byte sequences
+        .encode(Encoding.find(source_encoding), invalid: :replace, undef: :replace, replace: '') # Remove invalid byte sequences
         .gsub(/\r\r?\n?/, "\n") # Replaces windows line separators with "\n"
     end
 
@@ -75,5 +78,21 @@ def sanitize_cells(rows)
         end
       end
     end
+
+    def encode_cells(rows)
+      rows.map do |cells|
+        cells.map do |cell|
+          cell ? cell.encode(target_encoding) : ""
+        end
+      end
+    end
+
+    def source_encoding
+      encoding.split(':').first || 'UTF-8'
+    end
+
+    def target_encoding
+      encoding.split(':').last || 'UTF-8'
+    end
   end
 end