This research is about the development of a photocatalytic reactor design, Honeycomb, for in-situ groundwater remediation. Photocatalysis, typically a pseudo first order advanced oxidation process, is initiated via the illumination of UVA light on the catalyst, i.e. titanium dioxide (TiO2). In the presence of oxygen, highly reactive oxidising agents are generated such as superoxide (O2-), hydroxyl (OH·-) radicals, and holes (hvb+) on the catalyst surface which can oxidise a wide range of organic compounds. The target contaminant is methyl tert butyl ether (MTBE), a popular gasoline additive in the past three decades, which gives the water an unpleasant taste and odour at 20 µg L-1, making it undrinkable. This research consists of three major parts, i.e. (i) establishing a suitable catalyst immobilisation procedure, (ii) characterisation and evaluation of reactor models and (iii) scale up studies in a sand tank. TiO2 does not attach well onto many surfaces. Therefore, the first step was to determine a suitable immobilisation procedure by preparing TiO2 films using several potential procedures and testing them under the same conditions, at small scale. The coatings were evaluated in terms of photocatalytic activity and adhesion. The photocatalytic activity of the coatings was tested using methylene blue dye (MB), which is a photocatalytic indicator. A hybrid coating, which comprises a sol gel solution enriched with Aeroxide TiO2 P25 powder, on woven fibreglass exhibited the best adhesion and photocatalytic activity among samples evaluated. Thus, it was used to produce immobilised catalyst for this research. Consequently, the immobilisation procedure was scaled up to synthesize TiO2 coatings for the potential photocatalytic reactor design. The photocatalytic activity of the coatings produced from the scaled up immobilisation procedure were reasonably comparable to that produced at small scale. Due to the UVA irradiation and mass transfer limitations, photocatalytic reactors are typically compact in order to maximise their efficiency to accommodate high flows, particularly in water and wastewater treatment. In the case of groundwater, however, the treatment area can span up to meters in width and depth. Groundwater flow is significantly lower than that of water treatment, as the reactor design does not need to be compact. Considering both factors, a photocatalytic reactor design of hexagonal cross-section (Honeycomb) was proposed, in which the structures can be arranged adjacent to each other forming a honeycomb. A model was constructed and tested in a 4 L column (cylindrical) reactor, using the MB test to characterise the reactor performance and operating conditions. This was followed by a hydraulic performance study, which encompasses single and double pass flow studies. The single pass flow study involves the photocatalytic oxidation (PCO) of MB and MTBE, while the double pass flow study was focused on the PCO of MTBE only. The double pass can simulate two serially connected reactors. Single pass flow studies found that the critical hydraulic residence time (HRT) for the PCO of MB and MTBE is approximately 1 day, achieving up to 84 % MTBE removal. Critical HRT refers to the minimum average duration for a batch of contaminant remaining in the reactor in order to maintain the potential efficiency of the reactor. Double pass studies showed the reactor can achieve up to 95 % MTBE removal in 48 hours, and that reactor performance in the field of serially connected reactors can be estimated by sequential order of single pass removal efficiency. In groundwater, there are likely to be other impurities present and the effects of groundwater constituents on the reactor efficiency were studied. The MTBE PCO rate is affected by the presence of organic compounds and dissolved ions mainly due to the competition for hydroxyl radicals and the deactivation of catalyst surface via adsorption of the more strongly adsorbed organic molecules and ions. Despite the presence of organic compounds and dissolved ions, the reactor achieved about 80 % MTBE removal in 48 hours. A double pass flow study showed that the overall efficiency of the photocatalytic reactor in the field can be estimated via sequential order of its efficiency in a single pass flow study using the actual groundwater sample in the laboratory. A sand tank was designed for the simulation of the clean up of an MTBE plume from a point source leakage using the 200 mm i.d. Honeycomb I prototype. Honeycomb I achieved up to 88.1 % MTBE removal when the contaminated groundwater flowed through (single pass) at 14.6 cm d-1. The critical HRT for Honeycomb I was also approximately 1 day, similar to that in the column reactor. The response of MTBE removal efficiency towards flow obtained in the column reactor and sand tank was generic, indicating that the reactor efficiency can be obtained via testing of the model in the column reactor. The presence of toluene, ethylbenzene and o-xylene (TEo-X) decreased the MTBE removal efficiency in both the sand tank and column reactor. The same set of catalyst and 15 W Philips Cleo UVA fluorescent lamp was operated for a total of about 582 h (24 d) out of the cumulative 1039 h (43 d) sand tank experiments, achieving an overall MTBE removal efficiency of about 76.2 %. The experiments in the column reactor and sand tank exhibited the reliability of the immobilised catalyst produced in this research. This research demonstrates the potential of Honeycomb for in-situ groundwater remediation and also proposes its fabrication and installation options in the field.