Purpose <p>Individuals with specific fears, such as spider fear, have shown biases in attentional responding to fear-related stimuli as compared to those without fear. Research has linked these biases to the development of fear and to treatment outcomes. However, these studies have most often used pictographic stimuli and tasks with poor measurement reliability. A new method, the "dual-probe" video paradigm, has recently been developed to measure biased attention to naturalistic dynamic visual scenes with a high measurement reliability. The present study used the paradigm to examine the relationship between self-reported spider fear, attentional biases for spider stimuli, and choice preferences for viewing spider stimuli.</p> Methods <p>Non-select participants completed questionnaire measures of spider and butterfly fear, a “dual-probe” behavioural assessment of biased attentional responding to videos of spiders and butterflies in natural environments (Experiment 1 &amp; 2), and an assessment of choice preferences for viewing these videos (Experiment 2).</p> Results <p>Experiment 1 (N = 118) found that higher levels of spider fear predicted greater attentional avoidance of spider videos. Experiment 2 (N = 127) confirmed this result and showed that attentional avoidance was also predicted by a lower preference for viewing spider videos. Additionally, choice preferences statistically mediated the relationship between spider fear and attentional bias. Both the attention and choice assessment tasks demonstrated high measurement reliability (<i>r</i>_split-half &gt;.90).</p> Conclusions <p>This study shows that higher self-reported spider fear is characterised by attentional avoidance of naturalistic and dynamic spider stimuli, which can be explained in part by viewing preferences. It also demonstrates that the dual-probe video paradigm is a reliable method for measuring fear-related attentional biases and choice preferences.</p>