In the studies on Enterprise Systems (ES), different failure rates ranging from 50 to 90 percent are often reported. Despite these reported failure rates, ES market witnesses continuous growth. This paper presents a review of ES failure rates and critically analyses the numbers reported within their original context as well as in the context of various definitions of failure. We find that the reported high failure rates are historical, lack a consistent definition and lack an adequate understanding of ES outcomes. We find that ES outcomes are often complex, time-dependent, multi-dimensional and constructionist in nature. Based on this analysis, implications for ES research are discussed.