diff --git a/spec/report_spec.cr b/spec/report_spec.cr
index f9e9b64..97491c2 100644
--- a/spec/report_spec.cr
+++ b/spec/report_spec.cr
@@ -15,7 +15,7 @@ def new_failure_result(result_type : Spectator::Result.class = Spectator::Failed
   result_type.new(example, elapsed, expectations, error)
 end
 
-def new_report(successful_count = 5, failed_count = 5, error_count = 5, pending_count = 5, overhead_time = 1_000_000i64)
+def new_report(successful_count = 5, failed_count = 5, error_count = 5, pending_count = 5, overhead_time = 1_000_000i64, fail_blank = false)
   results = [] of Spectator::Result
   successful_count.times { results << new_passing_result }
   failed_count.times { results << new_failure_result }
@@ -24,7 +24,7 @@ def new_report(successful_count = 5, failed_count = 5, error_count = 5, pending_
 
   example_runtime = results.compact_map(&.as?(Spectator::FinishedResult)).sum(&.elapsed)
   total_runtime = example_runtime + Time::Span.new(nanoseconds: overhead_time)
-  Spectator::Report.new(results, total_runtime)
+  Spectator::Report.new(results, total_runtime, fail_blank: fail_blank)
 end
 
 describe Spectator::Report do
@@ -43,6 +43,13 @@ describe Spectator::Report do
     end
   end
 
+  describe "#examples_ran" do
+    it "is the number of non-skipped examples" do
+      report = new_report(5, 4, 3, 2)
+      report.examples_ran.should eq(12)
+    end
+  end
+
   describe "#successful_count" do
     it "is the expected value" do
       report = new_report(5, 4, 3, 2)
@@ -94,6 +101,31 @@ describe Spectator::Report do
         report.failed?.should be_false
       end
     end
+
+    context "with fail-blank enabled" do
+      context "when no tests run" do
+        it "is true" do
+          report = new_report(0, 0, 0, 5, fail_blank: true)
+          report.failed?.should be_true
+        end
+      end
+
+      context "when tests run" do
+        context "and there are failures" do
+          it "is true" do
+            report = new_report(5, 4, 3, 2, fail_blank: true)
+            report.failed?.should be_true
+          end
+        end
+
+        context "and there are no failures" do
+          it "is false" do
+            report = new_report(5, 0, 0, 2, fail_blank: true)
+            report.failed?.should be_false
+          end
+        end
+      end
+    end
   end
 
   describe "#remaining?" do
diff --git a/src/spectator/report.cr b/src/spectator/report.cr
index a284dff..11c4797 100644
--- a/src/spectator/report.cr
+++ b/src/spectator/report.cr
@@ -25,7 +25,8 @@ module Spectator
     # The *results* are from running the examples in the test suite.
     # The *runtime* is the total time it took to execute the suite.
     # The *remaining_count* is the number of tests skipped due to fail-fast.
-    def initialize(@results : Array(Result), @runtime, @remaining_count = 0)
+    # The *fail_blank* flag indicates whether it is a failure if there were no tests run.
+    def initialize(@results : Array(Result), @runtime, @remaining_count = 0, @fail_blank = false)
       @results.each do |result|
         case result
         when SuccessfulResult
@@ -46,9 +47,14 @@ module Spectator
       @results.size
     end
 
+    # Number of examples run (not skipped or pending).
+    def examples_ran
+      @successful_count + @failed_count
+    end
+
     # Indicates whether the test suite failed.
     def failed?
-      failed_count > 0
+      failed_count > 0 || (@fail_blank && examples_ran == 0)
     end
 
     # Indicates whether there were skipped tests
diff --git a/src/spectator/runner.cr b/src/spectator/runner.cr
index 41e78ee..27558e9 100644
--- a/src/spectator/runner.cr
+++ b/src/spectator/runner.cr
@@ -23,7 +23,7 @@ module Spectator
 
       # Generate a report and pass it along to the formatter.
       remaining = @suite.size - results.size
-      report = Report.new(results, elapsed, remaining)
+      report = Report.new(results, elapsed, remaining, @config.fail_blank?)
       @config.formatter.end_suite(report)
 
       !report.failed?