Neuropsychological evaluation may be of particular relevance in the detection of subtle cognitive impairments after mild traumatic brain injury (MTBI), including the subgroup of MTBI patients with a persistent postconcussion syndrome (PCS). Attention measures may be the most sensitive indicators of dysfunction associated with MTBI; however, previous studies have typically relied on the analysis of overall group differences, which may not reflect the diagnostic accuracy of attention measures when applied to individuals with MTBI. In the present study, subjects with persistent symptoms at least 3 months following a mild traumatic brain injury were compared with a sample of community living, normal control subjects in order to evaluate the sensitivity, specificity, and diagnostic accuracy of attention measures. Patients with PCS, screened with conservative inclusion and exclusion criteria, and a matched normal control group were administered six clinical tests of attention: Digit Span, Trail Making Test, Part A and Part B, Stroop Color-Word Test, Continuous Performance Test of Attention (CPTA), Paced Auditory Serial Addition Test (PASAT), and Ruff 2 & 7 Selective Attention Test. Consistent with prior research, these measures exhibited a wide range of sensitivity and specificity to possible cognitive impairment among patients. Attention measures may be the most sensitive indicators of dysfunction associated with PCS. Measures with high specificity (e.g., Stroop Color, and 2 & 7 Processing Speed) were shown to have strong positive predictive value, while measures with high sensitivity (e.g., CPTA) demonstrated strong negative predictive value for diagnosing PCS. Examination of the Odds Ratios indicated that measures assessing processing speed had a reliable, positive association with PCS, while measures without a processing speed component did not. Implications for making informed clinical decisions are discussed.