This study investigates linguistic cues to deception in written, real-life, high-stakes accounts in the computer-mediated communication context. Building on the theories of deception and reading, we developed hypotheses on the relationship between message veracity, linguistic cues to deception, reading patterns, and detection accuracy. We used a 250 Hz frequency eye tracker that records eye movement data every four milliseconds and allows the capture of eye movements in a remote, contact-free setup with free head movements. We used repeated measures mixed model in SPSS 24 to analyze the relationship between media and attention foci on linguistic cues to deception. The reading behavior of veracity judges varied across honest and deceptive statements. We also found support for the effect of visual foci on detection performance. In particular, fixation durations on generalizing words, on first-person pronouns, and fixation frequencies on passive voice verbs worsened detection accuracy. Fixation durations on passive voice verbs, third-person pronouns, and fixation frequencies on first-person pronouns, as well as the total fixation counts, improved detection rates.