Phosphate is a minor but important constituent of iron ores and is especially undesirable in the production of steel where it is considered a penalty element and is therefore tightly controlled. Iron ore quality must be determined at source, which means that deleterious elements must be accurately determined as an ongoing process. This demands fast, reliable analytical methods for the determination of, amongst others, phosphorus. However, such analysis is severely hampered by the following factors: the standard analytical method employed is X-ray fluorescence (XRF), but the measurements for phosphorus are seriously affected by matrix and inter-element interferences. An alternative technique is inductively coupled plasma with optical emission spectroscopy (ICP-OES), but there are also complications in using ICP-OES. The abundance of phosphorus in rocks is very low, generally under 0.5% wt. ICP-OES is also very insensitive to phosphorus and therefore must use the most sensitive spectral lines for quantification, and, there are serious spectral interferences on these lines by Fe and Cu. This study examines several analytical approaches for the determination of phosphate, using preparation by fusion and analysis by ICP-OES, demonstrating that suitable mathematical procedures and software applications can produce good results for phosphates, comparable to inter-laboratory tests or XRF.