Imaging flow in small channels for biochemical and biological applications has garnered increasing interest from the research community. Sensitive and accurate measurement and imaging technologies are required whether imaging blood flow in peripheral blood vessels or the flow of reagents in microfluidic circuits. In this context, optical sensing schemes are favourable as they are non-invasive, non-contact, chemically inert and have the potential for high spatial resolution and measurement accuracy. In the domain of microfluidics, micrometer-scale particle image velocimetry (µPIV) is the state-of-the-art for measuring flow, but is bulky, expensive, and requires tracer particles in the flow. A different technology, laser Doppler imaging, is simpler, more compact, and less expensive than µPIV, and has been used to image flow both in vivo and in vitro. Laser feedback interferometry (LFI) is a versatile sensing and imaging technique that can be configured to operate upon the same fundamental principle as conventional laser Doppler systems, but in a simpler form as it requires neither a reference arm nor an optical detector. LFI has been used to successfully measure blood flow in large (greater than 1 mm) flow channels and in vivo.In this dissertation, the practical and theoretical aspects of applying LFI to the measurement of flow in microchannels is investigated. A system utilising bulk external optics to focus laser light into a microfluidic channel is presented. The system is able to accurately measure the spatial flow profile within the channel for Newtonian and non-Newtonian flow. This system realisation gives rise to the practical question of how better to integrate the optics into the microfluidic system, as well as questions regarding the signal processing needed to extract the flow velocity, and the correct modelling of laser dynamics under this sensing scheme.To address the question of system integration, a practical realisation of an integrated optofluidic LFI system is presented. The system consists of an optical fiber embedded within a microfluidic circuit, and is shown to accurately measure flow over three decades of scatterer concentration, and two decades of fluid velocity.To improve signal processing, the application of the multiple signal classification (MUSIC) algorithm to LFI signals is investigated. It is first shown that for LFI velocimetry and absolute distance measurement systems, MUSIC exhibits superior performance in the presence of noise than the commonly used fast Fourier transform (FFT) method, and that for certain types of velocimetry measurements one may discard the ordinarily used fitting procedure. Following on from this, the application of MUSIC to the specific signals obtained from LFI flowmetry is investigated, showing MUSIC to have markedly better signal-to-noise ratio than the standard FFT method, allowing accurate and robust flow measurement at very low scatterer concentration, making it eminently suitable for microfluidic and nanofluidic systems.Finally, to address questions of modelling the complex laser dynamics of LFI systems measuring fluid flow and velocity of rough bodies in motion, a method to incorporate the frequency-shifted optical feedback with random phase into the Lang and Kobayashi rateequation model is presented, exhibiting good agreement between experiment and simulation in both time and frequency domains.These contributions together circumscribe a comprehensive investigation of LFI applied to flow measurement in small channels, addressing questions relating to practical, theoretical, modelling and signal processing aspects.