Dreams that appear to predict future events that could not have been anticipated through any known inferential processes have been reported for centuries, and dreams that appear to anticipate the death of an acquaintance or loved one are particularly common. Such reports become more suggestive of genuine precognition if there are no natural cues (such as an illness) to an impending death and if the time interval between the dream and the subsequent death is brief. Most reports are difficult to evaluate because we dream many times each night but typically remember and report only a salient subset of our dreams. Thus we cannot assess whether the time interval between a death-related dream and the death of the dream character is brief or lengthy because we have no control set of non-death-related dreams to which its time interval can be compared. The study reported here provides just such a control set by comparing deathrelated and non-death-related dreams featuring the same set of dream characters who died after the dreams occurred. These were drawn from the author's own dream journal in which he has recorded his nightly dreams for nearly twenty-five years. The mean time interval between death-related dreams and the person's subsequent death was significantly shorter than the time interval between non-death-related dreams and his or her death, t(11) = 3.30, p =.004, one-tailed. Cases in which death-related dreams occurred after the characters had died are also considered. Seven of the cases are discussed in detail.