Problem: While Cucumber is an awesome tool, for some kinds of tests it's default behaviour becomes an obstacle. Testing large batches of input against a slow error prone system is the source of much frustration. Cucumber will skip remaining steps on failure or error. This is especially problematic if the test input is a dynamic collection, *(e.g. results of a database query). This is opposed to a static collection testing which is solved by a Scenario Outline. Solution: We need to step putside Cucumber's default pass/fail/error handling. Capture all failures and errors in collections. Log errors and failures. Generate a pass/fail manifest against the test inputs. Do not use this for normal BDD style testing, Cucumber's default behaviour is perfect for that. Public Interface: This is a works in progress so I expect changes as usage reveals more. In your cucumber env.rb Before do |scenario| setup_unstoppable end After do |scenario| print unstoppable_failures(scenario) print unstoppable_errors(scenario) end In a step definition wrap any operation that you do not wish to stop execution like so unstoppable do expect(thing).to be(exected_thing) end This helper method does the following: runs executes the block catches any exception 2a. adds error to errors collection if an error 2b. adds expectation failure to failures collection if error is an RSpec::Expectations::ExpectationNotMetError logs error/failure