Kenneth Arrow first formally recognized the importance of organizational learning-by-doing in economics, observing that “it is the very activity of production which gives rise to problems for which favorable responses are selected over time” (Arrow 1962, p. 156).1 Empirical articles have established its existence across a broad range of manufacturing industries, but in healthcare industries, studies have focused primarily on scale effects rather than organizational learning-by-doing measured from when a provider first begins offering a procedure (e.g., Ho 2002; Gaynor et al. 2005). In addition, despite repeated evidence of the existence of organizational learning-by-doing and documentation of significant heterogeneity across industries (Balasubramanian and Lieberman 2010), only one existing paper considers intraindustry heterogeneity in organizational learning-by-doing (Pisano et al. 2001). Sources of inter-firm/intra-industry heterogeneity remain empirically unidentified in the literature. In liver transplantation, 6-month post-transplant survival rates increased from 64% in 1987 to 90% by 2009, indicating significant learning occurred during this time period. This paper seeks to identify whether learning-by-doing occurs at the level of the organization, whether it varies across organizations, and if the magnitude of learning-by-doing depends on how much uncertainty exists about how to perform the procedure at the industry level. As predicted by Arrow (1962) and in accordance with a broad literature in industrial organization, this study finds that organizational learning-by-doing exists in liver transplantation, but only very shortly after entry, approximately within the first 20 patients treated at a center. Extending a sixteen hospital study by Pisano et al. (2001), I find significant heterogeneity in the magnitude of learning-by-doing among the 124 centers in the sample, largely driven by the timing of entry. In accordance with theoretical predications (Jovanovic and Nyarko 1995), the amount of uncertainty at the industry level is a crucial factor in determining the extent of organizational learning-by-doing. In particular, the results show that evidence of learning-by-doing dissipates as liver transplantation advances over time from an experimental procedure to a common procedure with well-established protocols for care, specialized training programs, and better methods of immunosuppression. Organizational learning-bydoing also appears to be most important for mid-range survival, approximately 2–۳ months post-surgery. At 2–۳ months post-surgery, patient survival has shifted from being driven primarily by the surgical process to being determined by the careful balance of immunosuppression, enough to avoid organ rejection but not so much that the patient succumbs to fatal side effects or opportunistic infections. Although surgical expertise clearly is necessary, the limitations of and uncertainty about the optimal approach to immunosuppression always have been the primary constraints on the feasible duration of survival post-transplant. Correlations from other types of organ transplants support the association between the overall technical uncertainty in how to perform a procedure and the importance of organizational learning-by-doing.