Ambulance suspensions often give a poor ride which may result in deterioration in the condition of ill or injured patients. To reduce patient vibration, purpose-built ambulances or ambulances with modified chassis can be used. A potentially lower cost and more effective alternative is to provide additional isolation for the stretcher only. This thesis describes the design and performance of a pneumatic stretcher suspension which uses a novel linkage to provide isolation in bounce and pitch. Results of linear and non-linear analyses are presented which characterise the behaviour of the suspension. The kinematics of the suspension linkage are shown to give to a vertical stiffness which reduces in compression. Pneumatic cylinders connected to auxiliary tanks are used as springs. These give the suspension essentially load-independent natural frequencies of around 0.46 Hz in bounce and pitch. Damping is provided by fitting a flow restriction between the cylinders and tanks. By simulating the response of the stretcher to realistic random floor vibrations, it is concluded that a low level of damping is required and that an orifice restriction is preferable to a capillary restriction. These simulations are believed to be the first for which pneumatic damping is assessed by using a realistic random input. Additional simulation results demonstrate that improved isolation is possib1e by using an innovative semi-active pneumatic damper which is controlled according to the skyhook damping principle. A mechanical shaker which uses adjustable stroke round cams is described. Suspension tests carried out using this shaker are detailed. Various combinations of patient mass, pneumatic damping level, and shaker stroke and frequency are used. Acceleration transmissibilities are presented which indicate both that good levels of vertical isolation are obtained (eg. greater than 90% isolation above 5.2 Hz), and that isolation performance is largely independent of patient mass. Coulomb damping is shown to have a detrimental effect on isolation - particularly for low acceleration or high frequency inputs. The results of road tests are presented. These show that the suspension provides isolation above 1 Hz and reduces r.m.s. accelerations by 45-60%. The suspension is concluded to offer the potential to reduce patient vibration at reasonable cost, although improvements to the design are required in moving to a production model.