Supplier reliability affects numerous supply chain outcomes, including supplier market shares, inventory at different levels of a supply chain, and retailer cost and prices (Stank et al. 2003; Dada et al. 2007; Federgruen and Yang 2009; Liu et al. 2009; Davis-Sramek et al. 2010; Craig et al. 2016). Researchers typically model suppliers with imperfect product availability using the type 1 and type 2 inventory service level metrics. Type 1 service level, or in-stock, is the probability that a supplier will fill all demand in a given period. Type 2 service level, or fill rate, is the expected proportion of demand that a supplier will fill in a given period (Nahmias 2008). The literature on supplier reliability, however, offers a dynamic perspective on inventory service level not captured by traditional metrics. Researchers and practitioners often describe service level in terms of consistency, or predictability (Dana and Petruzzi 2001; Swait and Erdem 2002; Christopher 2005; Su and Zhang 2009; Solomon 2012). On the other hand, the literature also highlights the importance of a supply chain’s ability to recover from service disruptions (Bakshi and Kleindorfer 2007; Craighead et al. 2007; Sheffi 2007; Turner 2011). We extend prior research on retailers ordering from imperfect suppliers by incorporating aspects of supplier performance identi- fied in the literature, namely consistency (the ability of a supplier to fulfill orders repeatedly) and recovery (the ability of a supplier to fulfill orders after a lapse in service). We propose a stylized model of supplier service level that captures both consistency and recovery, and we develop a method for estimating consistency and recovery using data commonly available within supply chains. We demonstrate this method using data from a supplier of consumer packaged goods (CPG) as well as a supplier of apparel. Our results demonstrate the model’s ability to capture distinctions in each supplier’s performance, specifically differences between the consistency and recovery rates for each supplier. These differences reinforce the need to distinguish between these service dimensions. To further explore consistency and recovery, we model a retailer purchasing identical products from two suppliers. The model incorporates both the retailer’s uncertainty about its demand and the retailer’s uncertainty about its suppliers’ service levels. The retailer observes each supplier’s delivery history, updates its beliefs about each supplier’s consistency and recovery, and then places orders that minimize the expected market mediation costs (i.e., the expected cost of overages and underages). The retailer’s beliefs may be informal—as in a buyer’s opinion of a particular supplier—or formal— as in supplier scorecards (Duffy 2004). The model identifies how supply chain outcomes—in particular, market share across suppliers and retailer cost—vary with supplier performance. To study the impact of consistency and recovery in practice, we construct numerical experiments calibrated with empirical data. The numerical experiments reveal the extent to which a supplier’s orders from a retailer depend on the supplier’s consistency and recovery.