The hospitalization rate of HMOs is reported to be 444 bed days per 1,000 enrollees per year. It is often forgotten that there is also out-of-plan utilization. A review of previous studies and a survey of reporting practices by three HMOs illustrate many problems with HMO utilization data. HMO rates, like those of other insurers, reflect only the hospital admissions that the plans know about and pay for, not the total hospital experience of their enrollees. While only a thorough tracking of subscriber utilization of all insurers and institutions will provide estimates of the magnitude of unreported admissions and their impact on utilization rates, this report enumerates the ways in which patients may receive inpatient care without the HMO having a record of the admissions and/or having to pay for them. It was found that admissions can be unreported 1) when another insurer or institution pays (e.g., Medicare, No Fault, Workmen's Compensation, duplicate coverage, school health and liability insurance or VA, military, municipal, and state hospitals); 2) when the HMO does not cover benefits (e.g., cosmetic and oral surgery, experimental procedures, long-term psychiatric, chronic, or rehabilitation stays); and 3) when HMO coverage is denied for procedural reasons (e.g., catastrophic stays covered by reinsurance, newborns, voluntary “leakage”, or improper following of HMO procedures). True HMO rates are unknown but are estimated by some authors to be 7-37% higher than the reported figure, depending on which types of unreported use are estimated. There is a need for future analyses to quantify true hospitalization rates of enrollees of HMOs and other insurers.