Capacitive resistivity (CR) is a novel geophysical technique for the non-intrusive characterisation of the shallow subsurface by electrical imaging. CR is capable of extending the scope of the conventional DC resistivity technique to the urban built environment and other settings where galvanic contact cannot be achieved or where high contact impedances result in poor data quality. Fundamentally, the CR technique is based upon the concept of capacitive coupling between sensors and the ground and a generalisation of the DC four-point array for measuring the resistivity of the subsurface at frequencies in the VLF range. This thesis provides a unified description of CR, including its physical principles, their theoretical formulation and practical implementation in geophysical instruments. In general, the transfer impedance across a capacitive array is a complex function of frequency and geometry. It is shown that a low induction number mode of operation exists where resistivity is proportional to the in-phase component of the transfer impedance. The quadrature component is generally sensitive to a combination of parameters including sensor elevation, dipole offset and ground resistivity. Under the low induction number regime, the electric field is quasi-stationary so that theoretical equivalence with the DC case is achieved and conventional DC interpretation schemes are applicable to CR data. A comprehensive parameter study undertaken in this thesis investigates the applicability of the technique under the specific conditions typically encountered in environmental and engineering site investigation surveys. In those circumstances, practical CR measurements are shown to be limited to an optimal frequency window between 1 kHz and 25 kHz. The condition of low induction numbers imposes further restrictions on the maximum dimensions of the sensor array and the minimum resistivity of the ground. However, a key finding of the parameter study is that even under the quasi-static regime, practical conditions may be such that substantial phase rotations may occur which are exclusively due to the capacitive nature of the technique. Modelling of sensor capacitances is used to demonstrate that the concept of point poles postulated in the quasi-static formulation of CR has a practical realisation in the form of plate-wire sensors. Subsequently, the fundamental concepts of CR are validated experimentally in a series of elementary surveys where the fully complex transfer impedance (amplitude and phase) is measured with a newly developed prototype CR instrument. It is shown that for assessments of shallow subsurface conditions with typical survey parameters and standard geometries, the observed responses are typically in-phase. However, it is also demonstrated that practical circumstances exist under which significant phase rotations can be observed. In such cases, an estimation of apparent resistivity using the in-phase component only is more appropriate than the magnitude-based calculation performed by existing commercial instruments. The nature of the CR technique facilitates the use of towed arrays that allow the dynamic collection of multi-offset apparent resistivity data without the disadvantages of galvanic coupling. This thesis examines the operational characteristics of towed CR arrays and compares data acquired with a range of instruments in a variety of different environments. It is shown that towed-array CR enables the collection of highly repeatable resistivity data at sampling intervals of the order of centimetres. Towing-induced noise is found to be much less problematic than previously found with DC towed-array techniques. It is also demonstrated that high-quality data can be obtained by towed-array CR on artificial surfaces such as tarmac or concrete. Consequently, the technique appears to be particularly suited for assessing the condition of engineered structures such as roads and pavements. Finally, it is demonstrated how multi-offset towed-array CR can be employed for electrical tomographic imaging of the shallow subsurface. Conventional DC resistivity interpretation schemes based on quasi-2D, 2D and 3D inversion algorithms are shown to be applicable to such datasets, provided that some elementary rules are observed with regard to the design of towed-array surveys. Real-time interpretation during data acquisition is shown to be feasible with a continuous vertical electrical sounding (CVES) technique based on a Zohdy-type inversion. Examples of 2D and 3D surveys over shallow targets show the superior quality and resolution of CR datasets compared with conventional DC resistivity.