Tidal stream energy has the potential to contribute to global renewable energy generation, but this remains largely untapped. Technical developments have helped create a nascent industry with several pre-commercial installations. However, planned projects experience a cautious investment climate due to perceived risk of failure, and maintenance and repair cost. This paper reviews 58 tidal stream energy deployments between 2003 and August 2020. The analysis reviews commonalities, success and engineering issues, to inform current and future projects. The work classifies each deployment by type, rated power, number of devices, grid connection and foundation. In each case, project status and (if appropriate) failure mode is identified. Failure modes are compared to deployment classification to identify potential relationships. Most deployments were of horizontal axis turbines. The majority (54%) of deployments performed well. 18% failed, 14% were withdrawn from service, and 14% generated less power than planned. The most common failure cause was blade failure, followed by generator and monitoring failures. Ducted devices and devices in high velocity locations were more likely to fail, suggesting that flow velocity is a key factor. Most blade failures were attributed to underestimation of loads during design. Floating deployments were less likely to fail than fixed deployments, but more likely to be curtailed. Off-grid and grid connected deployments showed similar failure rates, suggesting sector immaturity. Tidal stream energy has accumulated around 1.4 million operating hours. Analysis shows a falling empirical failure rate, and likelihood of failure similar to that experienced by the wind industry at a similar stage. This work will be useful for project planners, developers and technology companies and investors in de-risking future project efforts.