Researchers have generally concluded that since the UCR data on murder and nonnegligent manslaughter have been within about 3-10 percent of the numbers listed in the Public Health Service category of 'homicide and legal intervention,' the UCR data are correct; however, based on the known justifiable homicide data, on the estimates by police in Detroit and elsewhere on excusable homicide, and on the substantial proportion of arrests and prosecutions which fail to result in conviction of criminal homicide, a reasonable estimate would be that 25-50 percent of reported criminal homicides are not in fact criminal homicides. Since the major criticism of the UCR data for other violent crimes is that they underestimate crime by about 50 percent, comparable reluctance should be expected where the UCR data may overestimate crime by as much as 50 percent. Since the number of well-disguised and undiscovered criminal homicides may equal the number of reported criminal homicides which are in fact criminal homicides, and thousands of noncriminal homicides are reported as criminal homicides, any analysis based on UCR data will be badly skewed. UCR data are also adequate for determining the involvement of particular weapons in criminal homicide. First, since it is not clear how many criminal homicides escape detection, there is no way to estimate the percentages of different weapons reported to and by the UCR. Second, neither the UCR nor its sources have an adequate basis for reporting what type of firearm was involved in gun-related homicide. Third, even if the handgun reports to the UCR were accurate, the data would still fail to indicate the involvement of handguns in criminal homicides, since a portion of the homicides reported by the UCR as murder or nonnegligent manslaughter are in fact excusable homicides. Further, neither the UCR nor the National Crime Survey can accurately indicate how many firearms-related crimes actually occur or what percentage of crimes of particular types involve firearms in general or handguns in particular. It is even less possible to determine the number of different firearms involved in those crimes. This study draws policy implications for data-collection improvements, analysis for policymaking bodies, and the use of UCR data as evidence. Thirty-two references are listed.