Introduction Malignant brain tumours constitute a small proportion of annual cancer incidence; however, the death rate is almost double the rate of incidence of all cases of cancer. This is not helped by the difficulties associated with the delivery of drugs to the brain. Steering magnetic nanoparticles (MNPs) i.e. superparamagnetic iron oxide nanoparticles (SPIONs) in an area of interest has been proposed for directing magnetically labelled drugs to clinical targets. A powerful magnetic and field gradient is required to ensure that the drugs are directed to the target area. In this case, the deeper the location, the stronger the magnetic force required. External permanent magnets can provide a strong magnetic field and gradient. We hypothesise that an inexpensive, portable, powerful and external Halbach array of 1.1T can be designed to steer SPIONs into brain tumours for drug delivery. Methodology Computational model: A 2-D simulation model of a Neodymium-iron-boron (FeNdB-52) magnet was run using Finite Element Method Magnetics (FEMM) software. In vitro model: The FEMM model was assembled and produced a Halbach array with a magnetic strength of 1.1T. The Halbach array was placed on the top of a 3D printed head in order to trap SPIONs (Fe3O4) in a 3D tumour, in contact with the surface of the head model. SPIONs were run through a fluid flow system using a syringe pump at 10 ml/min. The experiment was performed in two parts; the first part was to flow SPIONs with different concentrations ranging from 0.1mg/ml to 20mg/ml. The second part was to flow SPIONs, with the same concentrations, but loaded with 50 million white blood cells (WBCs) to mimic circulation. The concentration of SPIONs was quantified by inductively coupled plasma spectrometry (ICP). Images of the tumours with trapped MNPs were also obtained by MRI (3T) using a dual gradient echo sequence (TE= 4.60ms, 20ms). The uptake of SPIONs and viability of brain cells as well as brain tumour cells was evaluated using TEM and flow cytometry. Findings The strength of the modelled magnetic arrays was 1.6T whereas the actual assembled one was 1.1T assessed using Gaussmeter. The concentration of trapped SPIONs in the phantom flow model was different depending on the initial concentration of SPIONs and the location of the tumours within the head phantom. As expected, the further away the tumour was from the magnet, the less SPIONs were trapped. Trapping was confirmed by both ICP and MRI. The SPIONs uptake was observed in the cytoplasm of the cells and it was also noted that increasing the concentration of MNPs resulted in increased toxicity of the brain cells and brain tumour cells. This suggests that the concentration of iron needed to guide drugs with minimal toxicity should be considered. Conclusions Various Halbach array designs were modelled and an optimised design assembled. The in vitro experiments showed that the Halbach arrays could trap SPIONs with or without WBCs inside the tumour at different distances- up to 10 cm. This suggests that Halbach arrays have the potential to trap therapeutic drugs labelled with iron particles and we believe this would be useful for targeting of anti-cancer therapies to brain tumours.