The behaviour of large amplitude second mode internal solitary waves has been investigated. Waves were generated in the laboratory by a variety of methods and were observed using dye and particle visualisation techniques. Observations of wave generation, propagation and dissipation were used to develop new theories about the behaviour of such waves. Waves were generated by exchange flow, forced inflow and gravity collapse. The generation sequence was found to be similar for all three methods, with the gravity collapse technique allowing the most rapid and repeatable generation of second mode waves. External propagation characteristics (wave celerity and geometry) were investigated using waves of dimensionless amplitude up to a/h = 11.6. It was found that wave geometry was described well by existing theory over the entire range of amplitudes, but the existing wave celerity relationships were only accurate up to a/h ≌ 3. A new analytical approach produced a relationship which is applicable to all large amplitude waves. Internal propagation characteristics (internal circulation, entrainment and mass transport) were investigated using both particle visualisation and laser induced fluorescence techniques. It was found that internal circulation differs from the pattern suggested by existing numerical models. The interior of the wave is made up of an assemblage of vortices, symmetrical about the wave centreline, with a net flow rearwards along the centreline. These vortices are seen to play an important role in the entrainment of fluid into the wave. Entrainment appears to be caused by a non-symmetric Holmboe instability at the wave boundary. The entrainment into and expulsion of fluid from the wave results in the flushing of fluid from the wave. Measurements indicate that the rate of flushing is linearly proportional to distance and Richardson number during the primary flushing phase (from 100% to 10% tracer concentration). During the secondary flushing phase (from 10% to 1% tracer concentration) the flushing rate is lower but also linearly proportional to distance and Richardson number. Wave dissipation experiments indicate that wave amplitude decay rate is constant for any wave but varies with densimetric factor. Waves with larger densimetric factors decay at a slower rate. An expression for wave energy was formulated and the wave energy decay rate was examined. It was found that the radiation of first mode waves does not provide a significant contribution to wave decay. Wall shear was quantified and found to vary with flume width. In this study it was responsible for approximately 9% of the dissipation rate. The remaining dissipation is due to fluid drag (interfacial shear, pressure drag and mixing) and was quantified by a drag coefficient. The drag coefficient varies with the inverse of the cube root of densimetric factor.