diff --git a/app/controllers/bulkrax/guided_imports_controller.rb b/app/controllers/bulkrax/guided_imports_controller.rb
index 26f43e65..2bccd26e 100644
--- a/app/controllers/bulkrax/guided_imports_controller.rb
+++ b/app/controllers/bulkrax/guided_imports_controller.rb
@@ -39,7 +39,7 @@ def validate
       admin_set_id = params[:importer]&.[](:admin_set_id)
       validation_result = run_validation(csv_file, zip_file, admin_set_id: admin_set_id)
       raw_csv_data = validation_result.delete(:raw_csv_data)
-      cache_key = cache_validation_errors(validation_result, raw_csv_data)
+      cache_key = cache_validation_errors(validation_result, raw_csv_data, csv_file)
       formatted = StepperResponseFormatter.format(validation_result)
       formatted[:validationErrorsCacheKey] = cache_key
       render json: formatted, status: :ok
@@ -58,9 +58,10 @@ def download_validation_errors
       csv = ValidationErrorCsvBuilder.build(
         headers: cached[:headers],
         csv_data: cached[:csv_data],
-        row_errors: cached[:row_errors]
+        row_errors: cached[:row_errors],
+        file_errors: cached[:file_errors]
       )
-      send_data csv, filename: 'import_validation_errors.csv', type: 'text/csv', disposition: 'attachment'
+      send_data csv, filename: error_csv_filename(cached[:original_filename]), type: 'text/csv', disposition: 'attachment'
     end
 
     def create
@@ -105,13 +106,29 @@ def render_invalid_uploaded_files_response
     # @param zip_file [File, nil] an optional ZIP containing file attachments
     # @param admin_set_id [String, nil] optional admin set ID for validation context
     # @return [Hash] validation result data
-    def cache_validation_errors(validation_result, raw_csv_data)
-      return nil unless validation_result[:rowErrors]&.any?
+    def cache_validation_errors(validation_result, raw_csv_data, csv_file)
+      has_errors = validation_result[:rowErrors]&.any? ||
+                   validation_result[:missingRequired]&.any? ||
+                   validation_result[:unrecognized]&.any? ||
+                   validation_result[:emptyColumns]&.any? ||
+                   validation_result[:missingFiles]&.any?
+      return nil unless has_errors
 
       key = "guided_import_errors:#{session.id}:#{Time.now.to_i}"
       Rails.cache.write(
         key,
-        { headers: validation_result[:headers], csv_data: raw_csv_data, row_errors: validation_result[:rowErrors] },
+        {
+          headers: validation_result[:headers],
+          csv_data: raw_csv_data,
+          row_errors: validation_result[:rowErrors] || [],
+          file_errors: {
+            missing_required: validation_result[:missingRequired] || [],
+            unrecognized: validation_result[:unrecognized] || {},
+            empty_columns: validation_result[:emptyColumns] || [],
+            missing_files: validation_result[:missingFiles] || []
+          },
+          original_filename: filename_for(csv_file)
+        },
         expires_in: 1.hour
       )
       key
@@ -134,6 +151,13 @@ def apply_field_mapping
       @importer.field_mapping = Bulkrax.field_mappings['Bulkrax::CsvParser']
     end
 
+    def error_csv_filename(original_filename)
+      return 'import_errors.csv' if original_filename.blank?
+
+      base = File.basename(original_filename, '.*')
+      "#{base}_errors.csv"
+    end
+
     def set_locale_from_params
       I18n.locale = params[:locale] if params[:locale].present? && I18n.available_locales.include?(params[:locale].to_sym)
     end
diff --git a/app/services/bulkrax/validation_error_csv_builder.rb b/app/services/bulkrax/validation_error_csv_builder.rb
index f6a57474..744b49ad 100644
--- a/app/services/bulkrax/validation_error_csv_builder.rb
+++ b/app/services/bulkrax/validation_error_csv_builder.rb
@@ -3,44 +3,59 @@
 require 'csv'
 
 module Bulkrax
-  # Builds a CSV string containing only the rows from a validated CSV that have
-  # row-level errors. An `errors` column is prepended as column 1; multiple
-  # errors on the same row are joined with " | ".
+  # Builds a CSV string containing all validation errors from a guided import.
+  # File-level errors (missing required columns, unrecognized headers, empty
+  # columns, missing files) appear first as summary rows with a blank `row`
+  # cell. Row-level errors follow, one output row per errored data row.
   #
   # Usage:
   #   csv = Bulkrax::ValidationErrorCsvBuilder.build(
-  #     headers:    result[:headers],
-  #     csv_data:   result[:raw_csv_data],
-  #     row_errors: result[:rowErrors]
+  #     headers:     result[:headers],
+  #     csv_data:    result[:raw_csv_data],
+  #     row_errors:  result[:rowErrors],
+  #     file_errors: {
+  #       missing_required: result[:missingRequired],
+  #       unrecognized:     result[:unrecognized],
+  #       empty_columns:    result[:emptyColumns],
+  #       missing_files:    result[:missingFiles]
+  #     }
   #   )
   class ValidationErrorCsvBuilder
     # @param headers [Array<String>] original CSV headers in order
-    # @param csv_data [Array<Hash>] one entry per data row; each hash must have
-    #   :row_number (Integer, 1-indexed data row, so first data row == 2 matching
-    #   validator convention) and :raw_row (String-keyed hash of column=>value)
+    # @param csv_data [Array<Hash>] one entry per data row; each hash has
+    #   :raw_row (String-keyed hash of column=>value)
     # @param row_errors [Array<Hash>] each hash has :row (Integer) and :message (String)
+    # @param file_errors [Hash] file-level issues:
+    #   - :missing_required [Array<Hash>] each hash has :model and :field
+    #   - :unrecognized [Hash] column_name => suggestion_or_nil
+    #   - :empty_columns [Array<Integer>] 1-based column positions with no header
+    #   - :missing_files [Array<String>] filenames referenced but not found
     # @return [String] CSV content
-    def self.build(headers:, csv_data:, row_errors:)
-      new(headers: headers, csv_data: csv_data, row_errors: row_errors).build
+    def self.build(headers:, csv_data:, row_errors:, file_errors: {})
+      new(headers: headers, csv_data: csv_data, row_errors: row_errors, file_errors: file_errors).build
     end
 
-    def initialize(headers:, csv_data:, row_errors:)
+    def initialize(headers:, csv_data:, row_errors:, file_errors:)
       @headers    = headers
       @csv_data   = csv_data
       @row_errors = row_errors
+      @file_errors = file_errors
     end
 
     def build
       errors_by_row = group_errors_by_row
+      blank_data    = Array.new(@headers.length)
 
       CSV.generate(force_quotes: false) do |csv|
         csv << ['row', 'errors'] + @headers
 
+        file_level_error_rows.each do |message|
+          csv << [nil, message] + blank_data
+        end
+
         @csv_data.each_with_index do |record, index|
           row_number = index + 2 # header is row 1; first data row is row 2
-          next unless errors_by_row.key?(row_number)
-
-          error_messages = errors_by_row[row_number].map { |e| e[:message] }.join(' | ')
+          error_messages = errors_by_row[row_number]&.map { |e| e[:message] }&.join(' | ')
           raw_row = record[:raw_row] || {}
           csv << [row_number, error_messages] + @headers.map { |h| raw_row[h] }
         end
@@ -56,5 +71,29 @@ def group_errors_by_row
         hash[row_num] << error
       end
     end
+
+    def file_level_error_rows
+      messages = []
+
+      Array(@file_errors[:missing_required]).each do |entry|
+        messages << "Missing required column '#{entry[:field]}' (#{entry[:model]})"
+      end
+
+      Hash(@file_errors[:unrecognized]).each do |col, suggestion|
+        msg = "Unrecognized column '#{col}'"
+        msg += " (did you mean '#{suggestion}'?)" if suggestion.present?
+        messages << msg
+      end
+
+      Array(@file_errors[:empty_columns]).each do |pos|
+        messages << "Column #{pos + 2} has no header and will be ignored during import"
+      end
+
+      Array(@file_errors[:missing_files]).each do |filename|
+        messages << "Missing file: #{filename}"
+      end
+
+      messages
+    end
   end
 end
diff --git a/spec/services/bulkrax/validation_error_csv_builder_spec.rb b/spec/services/bulkrax/validation_error_csv_builder_spec.rb
index 010bc89a..22c9c0e2 100644
--- a/spec/services/bulkrax/validation_error_csv_builder_spec.rb
+++ b/spec/services/bulkrax/validation_error_csv_builder_spec.rb
@@ -19,10 +19,10 @@
         [{ row: 2, severity: 'error', category: 'missing_required_value', column: 'title', value: nil, message: "Required field 'title' is missing" }]
       end
 
-      it 'includes only the errored row plus the header' do
+      it 'includes all data rows plus the header' do
         result = described_class.build(headers: headers, csv_data: csv_data, row_errors: row_errors)
         rows = CSV.parse(result)
-        expect(rows.length).to eq(2)
+        expect(rows.length).to eq(csv_data.length + 1)
       end
 
       it 'puts the original row number in column 1' do
@@ -45,11 +45,12 @@
         expect(rows[1][4]).to eq('My Title')
       end
 
-      it 'excludes clean rows' do
+      it 'includes clean rows with a blank errors cell' do
         result = described_class.build(headers: headers, csv_data: csv_data, row_errors: row_errors)
         rows = CSV.parse(result)
-        source_ids = rows[1..].map { |r| r[3] }
-        expect(source_ids).not_to include('id-002')
+        clean_row = rows.find { |r| r[3] == 'id-002' }
+        expect(clean_row).not_to be_nil
+        expect(clean_row[1]).to be_nil
       end
     end
 
@@ -64,7 +65,7 @@
       it 'joins multiple error messages with " | "' do
         result = described_class.build(headers: headers, csv_data: csv_data, row_errors: row_errors)
         rows = CSV.parse(result)
-        expect(rows[1][1]).to eq('Title is required | Description is required')
+        expect(rows[3][1]).to eq('Title is required | Description is required')
       end
     end
 
@@ -76,17 +77,19 @@
         ]
       end
 
-      it 'includes one output row per errored input row' do
+      it 'includes all data rows' do
         result = described_class.build(headers: headers, csv_data: csv_data, row_errors: row_errors)
         rows = CSV.parse(result)
-        expect(rows.length).to eq(3) # header + 2 errored rows
+        expect(rows.length).to eq(csv_data.length + 1)
       end
 
-      it 'outputs errored rows in original order' do
+      it 'outputs rows in original order with errors on the correct rows' do
         result = described_class.build(headers: headers, csv_data: csv_data, row_errors: row_errors)
         rows = CSV.parse(result)
-        expect(rows[1][3]).to eq('id-001')
-        expect(rows[2][3]).to eq('id-003')
+        row_by_id = rows[1..].index_by { |r| r[3] }
+        expect(row_by_id['id-001'][1]).to eq('Duplicate source_identifier')
+        expect(row_by_id['id-003'][1]).to eq('Title is required')
+        expect(row_by_id['id-002'][1]).to be_nil
       end
     end
 
@@ -106,5 +109,80 @@
         expect(rows.first[2..]).to eq(headers)
       end
     end
+
+    context 'file-level errors' do
+      let(:row_errors) { [] }
+
+      it 'emits a row for each missing required column' do
+        result = described_class.build(
+          headers: headers, csv_data: csv_data, row_errors: row_errors,
+          file_errors: { missing_required: [{ model: 'GenericWork', field: 'title' }] }
+        )
+        rows = CSV.parse(result)
+        expect(rows[1][0]).to be_nil
+        expect(rows[1][1]).to eq("Missing required column 'title' (GenericWork)")
+      end
+
+      it 'emits a row for each unrecognized header, with suggestion when present' do
+        result = described_class.build(
+          headers: headers, csv_data: csv_data, row_errors: row_errors,
+          file_errors: { unrecognized: { 'legacy_id' => 'identifier', 'notes' => nil } }
+        )
+        rows = CSV.parse(result)
+        messages = rows[1..].map { |r| r[1] }
+        expect(messages).to include("Unrecognized column 'legacy_id' (did you mean 'identifier'?)")
+        expect(messages).to include("Unrecognized column 'notes'")
+      end
+
+      it 'emits a row for each empty column position' do
+        result = described_class.build(
+          headers: headers, csv_data: csv_data, row_errors: row_errors,
+          file_errors: { empty_columns: [3] }
+        )
+        rows = CSV.parse(result)
+        expect(rows[1][1]).to eq('Column 5 has no header and will be ignored during import')
+      end
+
+      it 'emits a row for each missing file' do
+        result = described_class.build(
+          headers: headers, csv_data: csv_data, row_errors: row_errors,
+          file_errors: { missing_files: ['photo.jpg'] }
+        )
+        rows = CSV.parse(result)
+        expect(rows[1][1]).to eq('Missing file: photo.jpg')
+      end
+
+      it 'leaves the row number blank for file-level rows' do
+        result = described_class.build(
+          headers: headers, csv_data: csv_data, row_errors: row_errors,
+          file_errors: { missing_files: ['photo.jpg'] }
+        )
+        rows = CSV.parse(result)
+        expect(rows[1][0]).to be_nil
+      end
+
+      it 'leaves data columns blank for file-level rows' do
+        result = described_class.build(
+          headers: headers, csv_data: csv_data, row_errors: row_errors,
+          file_errors: { missing_files: ['photo.jpg'] }
+        )
+        rows = CSV.parse(result)
+        expect(rows[1][2..]).to all(be_nil)
+      end
+
+      context 'when both file-level and row-level errors are present' do
+        let(:row_errors) { [{ row: 2, severity: 'error', category: 'test', column: 'title', value: nil, message: 'Row error' }] }
+
+        it 'outputs file-level rows before row-level rows' do
+          result = described_class.build(
+            headers: headers, csv_data: csv_data, row_errors: row_errors,
+            file_errors: { missing_required: [{ model: 'GenericWork', field: 'title' }] }
+          )
+          rows = CSV.parse(result)
+          expect(rows[1][0]).to be_nil   # file-level row has no row number
+          expect(rows[2][0]).to eq('2')  # row-level row has row number
+        end
+      end
+    end
   end
 end