A systematic review was conducted on the sensitivity of fish testing guidelines to detect the anti-androgenic activity of substances. Sequence Alignment to Predict Across Species Susceptibility (SeqAPASS) was used to investigate the conservation of the androgen receptor (AR) between humans and fish, and among fish species recommended in test guidelines. The AR is conserved between fish species and humans (i.e. ligand binding domain [LBD] homology ≥70%) and among the recommended fish species (LBD homology >85%). For model anti-androgens, we evaluated literature data on in vitro anti-androgenic activity in fish-specific receptor-based assays and changes in endpoints indicative of endocrine modulation from in vivo studies. Anti-androgenic activity was most consistently and reliably detected in in vitro and in vivo mechanistic studies with co-exposure to an androgen (spiggin in vitro assay, Rapid Androgen Disruption Activity Reporter [RADAR] Assay, and Androgenised Female Stickleback Screen). Regardless of study design (Fish Short-Term Reproduction Assay [FSTRA], Fish Sexual Development Test [FSDT], partial or full life-cycle tests), or endpoint (vitellogenin, secondary sexual characteristics, gonadal histopathology, sex ratio), there was no consistent evidence for detecting anti-androgenic activity in studies without androgen co-exposure, even for the most potent substances (while less potent substances may induce no (clear) response). Therefore, based on studies without androgen co-exposure (35 FSTRAs and 22 other studies), the other studies (including the FSDT) do not outperform the FSTRA for detecting potent anti-androgenic activity, which if suspected, would be best addressed with a RADAR assay. Overall, fish do not appear particularly sensitive to mammalian anti-androgens.