The effects of the key variables on the selective flotation of Mt. Weld phosphate ore were studied with the aim of optimizing the flotation process to yield a concentrate of the following specifications: >36% P 2O 5, <3% Fe 2O 3, P 2 O 5 ( Fe 2 O 3+ Al 2 O 3) >10 at a P 2O 5 recovery of >70%. The main difficulty was to reduce the iron content in the apatite concentrate to the level required. Analysis of the QEM∗SEM results for two batch flotation tests gave useful insights into the flotation behaviour of the various types of minerals, particles and composite particles present in the ore. It was found that iron in the concentrate occurred mainly as hematite and goethite, which were collected partly due to the incomplete depression of the liberated iron oxides and partly due to composite particles containing iron oxides. Systematic microflotation tests on purified samples of apatite and hematite provided the basis for understanding the mechanism of the flotation process. A key feature of the flotation was the use of sodium silicate as a selective depressant for iron oxides. The flotation pH, the modulus of sodium silicate, the pH and the time of standing of the stock solution of sodium silicate and the presence of polyvalent metal ions in the silicate solution all affected the selectivity with which the iron oxides were depressed. A final improved flotation flowsheet was proposed which achieved 81% P 2O 5 recovery at an iron contamination level of 2.7% Fe 2O 3 in the concentrate.