Although there are important exceptions, the literature on managed community care has had difficulty showing cost effectiveness. In this context, cost effectiveness is generally defined as substitution (avoiding hospital and nursing home expenditures) or as efficacy (achieving more positive morbidity and mortality outcomes). For the past several years, William Weissert and his colleagues have sought to increase cost effectiveness by developing a framework for ‘‘titrating’’ community care. They argue that if services were allocated on the basis of the potential to avoid undesirable outcomes, then more resources would be devoted to those with greater risk and greater average benefit would result. Using data from the Arizona Long Term Care System (ALTCS), Weissert and his colleagues have explored the heterogeneity of chronically ill elder program participants with respect to mortality, hospitalization, nursing home use, and functional status. They have made the following arguments. First, individual long-term care participants have differential risks for these outcomes that can be reasonably estimated from assessment data (Chernew, Weissert, & Hirth, 2001). Second, although the community-care literature offers at best inconclusive evidence for the benefits of services in reducing these risks, it is possible to assert that services have small but consistent (invariant with respect to risk level) impacts on reducing risks in the four areas (Miller & Weissert, 2000). Third, the benefits of avoiding each of the four outcomes can be expressed in dollar terms (that are again invariant with respect to risk level or other factors; see Weissert et al., 2001). The three elements (risk, impact of services on reducing risk, and benefit in dollars) can be combined to estimate the relative value of community care for individuals, guide resource allocation, or establish reimbursement levels. Weissert and colleagues acknowledge the host of data limitations and technical concerns that limit the precision and generality of the effectiveness and risk-weighted value (ERV) estimates they have derived from the Arizona program experience. Nonetheless, they assert the potential value of ERV concepts and information in redirecting the attention of community-care program managers and practitioners. This may be too large a leap because some of their underlying assumptions may not survive focused tests. In particular, it seems unlikely that the value of community care in reducing the four risks is invariant across risk levels or insensitive to the particular combinations of services and supports allocated. Similarly, it seems unlikely that various outcomes are evaluated in the same ways by individuals from diverse social contexts or at diverse phases in the disease and disablement process. In this issue, Weissert and colleagues extend their ideas to care-planning decisions by ALTCS case managers (Weissert, Hirth, Chernew, Diwan, & Kim, 2003). They taught casemanagers ERV concepts and offered them information about anticipated risks and relative potential benefits of care (in relation to mortality, nursing home and hospital use, and functional status) for a varied group of hypothetical cases. Compared with case managers given only assessment data, those exposed to ERV rankings developed care plans that were on average less costly, devoted more resources to medical care, and were more closely correlated with risk ratings. Although these results support the conclusion that caremanagement choices can be influenced by an explicit resource-allocation paradigm and associated information, it should also be noted that ALTCS practitioners work in a capitated managed-care setting and thus face different care-planning opportunities, constraints, and incentives than those in other settings. It is yet to be seen, as Weissert and colleagues note, whether ERV-influenced resource-allocation choices would be sustained in real-world practice (with more potential influence on care plans of personal interactions with individual clients-caregivers). Furthermore, even if new allocation patterns were achieved, it is yet to be demonstrated that actual outcome patterns would shift in the hypothesized ways. Beyond these specific questions about the impacts of implementing the ERV paradigm on real-world Address correspondence to John A. Capitman, PhD, Schneider Institute for Health Policy, Brandeis University Heller School, 415 South Street, Waltham, MA 02454. E-mail: capitman@brandeis.edu Brandeis University Heller School, Waltham, MA.