Female partners of prostate cancer (PCa) survivors experience heightened psychological distress that may be greater than that expressed by PCa patients. However, optimal approaches to detect distressed, or at risk of distress, partners are unclear. This study applied receiver operating characteristics analysis to evaluate diagnostic accuracy, sensitivity and specificity of the Distress Thermometer (DT) compared to widely used measures of general (Hospital Anxiety and Depression Scale) and cancer‐specific (Impact of Events Scale‐Revised) distress. Participants were partners of men with localised PCa (recruited around diagnosis) about to undergo or had received surgical treatment (N = 189), and partners of men diagnosed with PCa who were 2–4 years post‐treatment (N = 460). In both studies, diagnostic utility of the DT overall was not optimal. Although area under the curve scores were acceptable (ranges: 0.71–0.92 and 0.83–0.94 for general and cancer‐specific distress, respectively), sensitivity, specificity and optimal DT cut‐offs for partner distress varied for general (range: ≥2 to ≥5) and cancer‐specific (range: ≥3 to ≥5) distress both across time and between cohorts. Thus, it is difficult to draw firm conclusions about the diagnostic capabilities of the DT for partners or recommend its use in this population. More comprehensive screening measures may be needed to detect partners needing psychological intervention.