The sound transmission loss of a range of plywood panels was measured to investigate the influence of the orthotropic stiffness of the plywood panels. The plywood panels were tested as single and also double leaf partitions, with a range of stud configurations. A new method was developed for predicting the sound transmission loss of single leaf partitions with both orthotropic and frequency dependent stiffness values. The sound transmission loss was evaluated for two significantly different sample sizes. The observed influence of the sample size on the measured sound transmission loss was profound. The construction of the partition was shown to significantly affect the influence of the sample size on the sound transmission loss. A qualitative analysis based on existing published research of the contributing factors is presented, and methods for adjusting the results for the small sample size for comparison with the large results were developed. The influence of a range of acoustic treatments of lightweight plywood partitions was investigated. The treatments involved internal viscoelastic materials and decoupled mass loaded barriers in various arrangements. The attachment between the treatment and the plywood panel was found to influence the sound transmission loss significantly. A prediction method based on published models was modified to allow the influence of the treatments to be included. Reasonable agreement was achieved between the predicted and measured results for a wide range of samples. A prediction method was developed that accounts for the influence of orthotropic, frequency dependent material parameters. This method utilised an adaptive, numerical integration method to solve an analytical formulation for the sound transmission loss. The influence of the finite sample size was accounted for using an expression for the finite panel radiation impedance. The finite panel radiation impedance was predicted analytically and an approximation was also presented. The presence of a significant source room niche was accounted for by applying an appropriate limit to the integration range of the angle of incidence. The prediction methods developed are compared with the measured transmission loss results from both the small and large test facilities. Good agreement was seen for some of the predicted results. Generally the agreement within the coincidence region was worse than for the rest of the transmission loss curve. The inclusion of orthotropic and frequency dependent stiffness values significantly improved the agreement within the coincidence region.