To explore the diagnostic value of a galactomannan (GM) detection for non-immunocompromised critically ill patients with influenza-associated aspergillosis (IAA). In this retrospective case-control study, we explored the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic (ROC) curve (AUC) of serum and bronchoalveolar lavage fluid (BALF) GM tests by four detection strategies at different detection time points and with different compound modes. In total, 90 patients were evaluated. The AUC values of the second serum GM test, the first and second BALF GM tests, were significantly higher (0.839 (95% CI 0.716 to 0.963), P < 0.01; 0.904 (95% CI 0.820 to 0.988), P < 0.01; 0.827 (95% CI 0.694 to 0.961), P = 0.043) than that of the first serum GM test (0.548 (95% CI 0.377 to 0.718)). We found that at least one positive result on two consecutive serum GM tests (0.719 (95% CI 0.588 to 0.849)) was the best compared with the first positive test (0.419 (95% CI 0.342 to 0.641), P < 0.01) and positives on two consecutive tests (0.636 (95% CI 0.483 to 0.790), P = 0.014). However, there were no differences between those three detection strategies of BALF GM. The BALF GM test might have a better diagnostic value for IAA in the ICU than the serum GM test. A possible cutoff value of 1.0 to 1.3 was set for GM from BALF specimens for IAA. A single serum GM test is not routinely recommended, but at least one positive result on two consecutive tests appeared to be useful.