The hydrodynamic effects of long-line mussel farms are studied through a two-pronged approach. Large-scale hydrodynamic effects are investigated through the use of field measurements, primarily at a large mussel farm in Golden Bay, New Zealand (230 long-lines, covering an area of 2.45 km by 0.65 km). The research focuses on three areas: the effect of the farm on currents, mixing and stratification, and the dissipation of wave energy. Measurements are also made of the forces on long-line anchor ropes, and a limited investigation is made of phytoplankton depletion. The second approach is the use of laboratory drag measurements and Particle Tracking Velocimetry (PTV) to study the effect of mussel dropper (vertical lengths of mussel-encrusted crop rope) roughness and spacing on flow at small scales. These experiments provide data on very rough cylinders, and on cylinder arrays. The field measurements show that the local effects of mussel farms on currents are significant, but that magnitudes of the effects depend on dropper density, mussel sizes, orientation of the long-lines to the flow, and other parameters that are necessary to characterise the complex interactions between a farm and the flow. The drag on the submerged structures reduces water velocities within the Golden Bay farm by between 47% and 67%. Mussel farms present a porous obstacle to the flow, and flow that does not pass through the farm must be directed around or beneath it. The field measurements indicate that at the study site, most of the flow is diverted around the farm despite its large horizontal dimensions. The droppers at the study site extend over most of the water column (average dropper length ~ 8 m, average water depth ~ 11 m), providing a restriction to the flow beneath the farm. The strength of the density stratification may also favour a horizontal diversion. The flow around the farm is essentially two-dimensional. This suggests that two-dimensional numerical models should be sufficient to obtain reasonable predictions of the velocity drop within, and the diversion around, mussel farms. A simple two-dimensional pipe-network model gives reasonable estimates of the velocity within the farm, demonstrating that the drag of the farm may be adequately parameterised through local increases of bed friction. A wake in the form of reduced velocities extends downstream of the farm, and a mixing layer analogy suggests that this wake spreads slowly. The downstream extent of the wake cannot be determined, although it is likely to be limited by the tidal excursion. The degree of vertical mixing caused by the flow through a mussel farm cannot be quantified, although there are clear interactions between the stratification and the farm. Two mixing mechanisms are considered. A shear layer is generated beneath the farm due to the difference in velocities between the retarded flow within the farm and the flow beneath. Shear layers beneath mussel farms are likely to be weak unless the ambient currents are strong. It will be necessary for stratification to be weak or non-existent for this mechanism to generate significant mixing. The second mechanism is smaller-scale turbulence generated by the mussel droppers. Although the efficiency of this form of mixing is likely to be low, the large number of mussel droppers suggests that there will be some enhancement of vertical mixing. Frequency-dependent wave attenuation is recorded, and is predicted with some success by an analytical model. Both the model and the field data show that wave dissipation increases as the wave period decreases. Wave energy dissipation at the study site averages approximately 10%, although the measurements are made during a period of low wave heights (Hs < 0.25 m). Measurements of long-line anchor rope tension at two study sites indicate that the loadings are induced by the tide, currents, and waves. Dynamic wave loadings may be significant, and higher wave forces are measured at the offshore end of a long-line. The issue of seston or phytoplankton depletion is considered briefly through the examination of fluorescence, turbidity, and acoustic backscatter data. Although the results are consistent with a reduction of seston within the farm, differences between the inside and outside of the farm are not statistically significant. Mussel droppers resemble extremely rough circular cylinders, with the mussel shells forming the surface roughness elements. Drag measurements and PTV flow visualisation are used to investigate the importance of the large surface roughness, and the influence of dropper spacing and long-line orientation on flow. Drag measurements conducted with smooth and rough cylinders show that high surface roughness (ks/D ~ 0.092) has little effect on the drag coefficient of single cylinders in the range 4,000 < Re < 13,000, yet increases the drag coefficient of a row of cylinders normal to the flow. High surface roughness on single cylinders has the effect of shortening the near-wake region, increasing the peak turbulent kinetic energy (TKE) behind the cylinder, and decreasing the Strouhal number (St = 0.21, 0.19, 0.17 for ks/D = 0, 0.048, and 0.094 respectively). Arrays of rough cylinders (ks/D = 0.094) demonstrate similar flow characteristics to those of smooth cylinders. At cylinder spacings of S/D < 2.2, the surface roughness acts to favour the formation of a particular metastable wake pattern, whereas different metastable wake patterns are formed each run behind the smooth cylinders. The experiments show that the drag on single row arrays of cylinders are related to the cylinder spacing (increasing drag with decreasing spacing), and the drag also varies with the sine of the angle to the flow, except where the array is at low angles to the flow. The PTV measurements provide new data regarding the two-dimensional distributions of velocity, TKE, and turbulence statistics behind the cylinder arrays.