A 70-day experiment was implemented to study the iron (Fe) bioavailability of three Fe sources in Epinephelus coioides (initial weight, 12.19 AE 0.14 g). Treatments consisted of 0, 50, 90, 130, 170 and 210 mg supplemental Fe/kg from iron sulphate (FeSO 4 ), ferric citrate (Fe-citrate) and hydroxy methionine analogue iron (Fe-MHA). The results showed that the growth performance was not affected by Fe level regardless of Fe sources. Hepatic catalase activity (CAT), the haematocrit value (Hct), haemoglobin (Hb), mean corpuscular volume (MCV) and Fe content in whole body or liver firstly increased significantly and then plateaued as Fe level increased regardless of the Fe source, except significant decreasing of Hct in 210 mg Fe/kg group from FeSO 4 . Manganese (Mn) content decreased significantly and then kept stable in the whole body and liver whatever the source was. The slope-ratio method using FeSO 4 as the reference and Hb content, CAT activity, body and hepatic Fe concentration as the response parameter showed the average relative bioavailability of Fe-citrate and Fe-MHA was 103.25 AE 4.35 (n = 4) and 134.5 AE 15.67 (n = 4), respectively, with the maximum value of Fe-MHA in any case. In conclusion, Fe-MHA was the most effective to meet preferable haematological index, CAT activity and mineral deposition, with little difference between Fe-citrate and FeSO 4 , athough Fe levels with different sources failed to influence the growth in grouper.