Cognitive models of anxiety vulnerability and psychopathology have proposed that individual differences in anxiety are characterised by biases in attentional responding to threat valenced information. Traditional approaches used to examine these biases have measured selective attention to stimuli that represent threats, such as pictures of negative scenes, angry faces, or feared animals. Though these approaches have repeatedly yielded evidence of these effects the measurement reliability of these approaches has been poor, limiting the confidence with which hypotheses can be tested. Recent research has demonstrated that biased attention to cues signalling imminent threat can be measured with acceptable reliability, suggesting a credible avenue for investigating anxiety-linked differences in attentional processing of threat valenced information. However, the relation between this process and anxiety vulnerability has not been tested. The present study conducted three experiments that recruited participants who varied in trait and state anxiety (N = 134), social anxiety (N = 122), or spider fear (N = 131) to complete an assessment of selective attention to cues indicating the location of threat valenced stimuli. Analyses revealed each anxiety domain to be associated with attentional vigilance, or avoidance, of threat cues and the reliability of selective attention indices to be acceptable (r-split-half = .69 - .81). The findings demonstrate multiple anxiety domains are characterised by biased attentional responding to cues signalling threat and that such biases can be measured reliably. Implications for theories of anxiety-linked attention processing and further avenues for investigating the bias, and its relation to anxiety vulnerability, are discussed.