Polymethylmethacrylate (PMMA) is used in many applications requiring good optical properties, resistance to ageing and resistance to UV irradiation, in spite of its relatively poor mechanical response to impact loading. In order to extend the range of application of this transparent but fragile material, and improve its competitiveness with engineering polymers such as polycarbonate, it is necessary to improve its fracture resistance. Several generations of toughened PMMA reinforced by addition of rubber particles have therefore been developed by Arkema over the last decade ("rubber toughened PMMA" (RTPMMA)). However, there remains a need for substantial improvement even in these relatively fracture resistant materials, providing the overall motivation for this project. The first type of material considered is based on RTPMMA in which the matrix ductility has been improved by copolymerization with up to 25 % ethyl acrylate (EA). The reinforcing particles used in these new modified RTPMMA grades comprise a PMMA core, either one (conventional 3L particles), or two rubbery layers (experimental 4L particles), and finally an outer grafted PMMA shell. A second, entirely new type of rubber modified PMMA, based on Poly(MMABA-MMA) copolymers (MAM) prepared by controlled radical polymerization (PRC) is then considered. PRC is carried out using nitroxide initiators, making it possible to polymerize the blocks of PMMA from the blocks of PBA in a continuous mass process. Although little was known about their mechanical response prior to the present work, the MAM copolymers represent an attractive alternative to RTPMMA from an economic point of view since their production involves essentially only one step. The scientific aim of this study is to understand the relationship between the mechanical properties, the crack propagation mechanisms and the microstructure of these various impact reinforced PMMA materials, and, if possible, to use this knowledge to identify strategies for improving their performance. It is first shown that there is an increase in the strain to break from 10 % to up to 30 % on addition of 25 % EA to the PMMA matrix, and a corresponding reduction in the yield stress from 95 to 60 MPa. The TEM observations of thin films deformed in-situ show these changes to be associated with a change in microdeformation mechanism from crazing in the pure PMMA to homogeneous plastic deformation. However, they are not accompanied by any substantial improvement in the high speed fracture resistance, the value of the maximum stress intensity factor, KImax, at 1 m/s remaining at about 2 MPa.m1/2 in all the matrices investigated. The ductility improves significantly on addition of the 3L and 4L particles, and this improvement is particularly marked with the 4L particles, the strain to break reaching 80 % and the yield stress dropping to about 40 MPa, regardless of the matrix EA content. The fracture resistance also increases over the full range of speeds investigated, and optimum performance is obtained in the matrix containing 25 % EA combined with 4L particles, KImax ranging from 3.2 MPa.m1/2 at 10-4 m/s to 3.5 MPa.m1/2 at 1 m/s. At low speeds, both the 3L and 4L particles deform by cavitation of the rubbery layer and, as the EA content is raised, the matrix microdeformation mechanism at the crack tip changes from crazing to simple shear, consistent with the observations for the matrix. Nevertheless, at very high speeds, at which crazing dominates in all the materials, the rubbery phase of the particles is argued to become relatively rigid and hence increasingly ineffective in decreasing the crack propagation energy. Thus, the overall improvements in mechanical behaviour obtained on EA addition are not reflected by an improvement in the notched Charpy impact toughness, which does not exceed 5.5 kJ/m2 in any of the RTPMMA. There may be scope for further optimizing the particle size at high EA contents owing to an observed decrease in entanglement molar mass, as demonstrated in previous work by Béguelin et al., but this is not thought to be viable from an economic point of view. MAM copolymers provide an alternative, low cost means of introducing a rubbery phase with controlled dimensions and morphology into a PMMA matrix. These MAM copolymers show similar ductility in tension to the RTPMMA (strain to break around 70 %) and an even lower yield stress (around 25 MPa depending on the grade). MAM is found to show a lamellar morphology for PMMA contents, ∫PMMA < 75%, and deformation of this morphology results in extension of the PBA phase and formation of chevron-like regions of shear deformation. Even so, the fracture resistance remains relatively low in the first generation of these materials, KImax not exceeding 2 MPa.m1/2 at 1 m/s, and Charpy impact tests on notched samples showing relatively little improvement between MAM and the PMMA homopolymer (notched Charpy impact toughness of up to 2.25 kJ/m2 compared with 1 kJ/m2 for PMMA and 5.5 kJ/m2 for RTPMMA). The morphological observations and mechanical test results for MAM nevertheless point to the important role of the molar mass and ∫PMMA, and also to the interest of stripping, i.e. removal of residual BA monomer prior to polymerization of the MMA block. It has hence been possible to design and produce a second generation of stripped MAM copolymers with higher molar masses (Mn > 100 kg/mol) and an optimum ∫PMMA (< 65 %). These are found not only to show ductile behaviour in tension and excellent optical properties, but also notched Charpy impact toughnesses approaching 6 kJ/m2. Although this is still significantly less than for polycarbonate (12 kJ/m2), it demonstrates the promise of MAM for expanding the range of applications of MMA-based systems beyond that currently accessible to conventional RTPMMA.