This study investigated the role of need for closure (NFC) and anxiety in delusions. The Need For Closure Scale (NFCS) and measures of trait and state anxiety were administered to an early psychosis group with current delusions, a clinical (generalised anxiety disorder--GAD) control group and a non-clinical control group. The battery of questionnaires was repeated at follow-up one year later. The NFCS did not meet criteria for a unidimensional scale and two sub-scales were removed from all further analyses. At baseline the deluded and GAD groups scored significantly higher on the reduced NFCS (NFCS-R) than the control group. Trait anxiety was related to NFCS-R in the GAD and non-clinical control groups, but not in the deluded group. At follow-up all groups scored significantly lower on the NFCS-R, perhaps suggesting a practice effect, although the two clinical groups continued to have higher scores than the non-clinical control group. The recovered and non-recovered deluded groups did not differ on the NFCS-R one year later, unlike in the GAD group where recovered participants had significantly lower scores than the non-recovered. Change scores on the NFCS-R and trait anxiety were correlated at trend level in the GAD and non-clinical control groups, but not in the deluded group. These findings suggest that whilst NFC and trait anxiety are related in non-psychotic groups, NFC may be implicated in the formation of delusions, independently of anxiety, in psychotic individuals.