Thick-film technology has found applications on miniaturised hybrid circuits in various fields (automotive electronics, televisions, ...). This technology is also now widely used for the fabrication of force and pressure sensors that use the piezoresistive properties of thick-film resistors. The goal of this work has been generated by the fact that usual piezoresistive pastes / inks were optimised for applications on alumina, which is the standard substrate for thick-film technology, but ill suited for more flexible substrates such as aluminium, steel or Ti alloys. We were limited by the process conditions of the commercial pastes, in particular the too high firing process that does not allow the use of substrates with melting temperature < 850°C. This technological lack leads to manufacture a new generation of piezoresistive pastes with low firing temperatures (Tf: 500 ... 700°C). In parallel, we aim to optimise the electrical properties (resistance R, temperature coefficient of resistance TCR and gauge factor GF values) by highlighting the link with the structural evolution during the firing process and the obtained properties, and by understanding the conduction process in such percolative systems. Study of usual commercial piezoresistive pastes allowed us to determine that such piezoresistive pastes are composed of a percolating network of nanoconductive RuO2 grains embedded in a lead borosilicate glassy matrix. Evolution during the firing process was emphasised and showed the importance of controlling the firing parameters to assure the best properties for the final thick-film. Commercial pastes are characterised by a TCR value close to 0 ppm/°C, a reasonable sheet resistance value (R ~ 10 kOhms) and a gauge factor comprise between 10-12, that can be influenced by structural and process parameters. Indeed, complementary studies on sensitivity and stability were realised, because of limited available information in literature concerning the effect of firing schedule, particularly of quenching, and have shown that these properties are very dependent on the conditions of firing, although the main commercial pastes showed a moderate stability. In fact, this study showed that a compromise should be found between the different properties (for instance, high GF pastes presents a poor stability), and emphasises the fact that they should be optimised. A manufacturing process has been developed, process never well described in the literature, leading to the realisation of different lead borosilicate glasses. It has resulted in the ability to realise three series of model piezoresistive pastes with different ranges of firing temperatures corresponding to high (700°C), low (600°C) and very low (500°C) firing temperatures. The control of several parameters (glass composition, conductive phase concentration, grain size, firing temperature...) allowed us to direct precisely our research to elucidate the principle of conduction in such percolative systems and the reactions occurring between the elements and their influence on the electrical properties. Structural and electrical properties were studied by varying diverse parameters such as conductive grain size, concentration and firing temperature, and a coherence was found between the electrical behaviour (conduction process) and its relation to the complex nanostructure. In other words, this key chapter presents the results and their interpretation by a model of conduction based on a nonuniversal tunnelling percolation theory and based on a previously unpublished hypothesis. Indeed, it was demonstrated that the piezoresistive response of the pastes changed dramatically depending on whether the composites were universal or not. For the composites with critical exponent t ~ 2, the piezoresistive factor Γ showed no dependence upon the RuO2 volume fraction x, whereas the nonuniversal composites displayed a logarithmic divergence of Γ near the percolation threshold. We have interpreted the piezoresistivity results as being due to a strain dependence of the critical exponent when this was nonuniversal. We have brought forth a microscopic formulation to the phenomenological level proposed by Balberg, and we can now assert that thick-film resistors (TFR) are mainly nonuniversal compounds showing transport exponent t larger than the universal limit t = 2.0. This exponent t depends on strain and leads to a logarithmic divergence of the gauge factor. The possibility of influencing t by external means (e. g. strain) has never been studied so far. We have proposed a new way to investigate percolative systems by studying the behaviour of piezoresistive pastes. After having elucidated the conduction mechanism in such piezoresistive pastes, we studied the influence of different parameters (Tf, grain size, concentration, dwell time) on the main electrical properties (R, TCR and GF). Structural analysis gave a possible interpretation of the results. RuO2 parameters have direct effects on the R, TCR and GF values. Tf acts on microstructure provoking interactions between the bulk components and the substrate (in case of high Tf), and consequently leading to a modification of the electrical properties. The same complementary studies as commercial pastes on stability showed a combined influence of the cooling rate and the temperature dwell-time on R and TCR values. The results are in coherence with commercial pastes. The evolution of the values can be explained by diffusion phenomenon and local microscopic strains due to important cooling rates. The evolution of R upon annealing 250°C was found to depend strongly on the cooling rate for commercial and model pastes, but this observed trend tends to saturate. These new series of low firing temperature were shown to be not as stable as the "best" commercial pastes, but their variations are much similar to "medium" commercial one's. At 250°C, possible evolution mechanisms could involve Ru in glass (dissolved or in clusters), or mechanical relaxation that can be extrinsic (macroscopic thermal mismatch between resistor and substrate) or intrinsic (local thermal mismatch between glass and conductive phase), and which can later relax during annealing. During this analysis, technological problems have been emphasised and a section was dedicated to resolve the problem of the unsuitability of the substrate to the very low firing temperature system, which showed local strain that induced cracks and leading to electrical instability. Moreover, it was shown that these new pastes could be optimised by additives or used on more adapted substrates. However, these obtained series offers a large range of TCR and R values for different low Tf and it would be useful for technological goals. The best proof of the success of our study was the realisation of sensor prototypes based on different substrates such as steel, aluminium and even glass. This work has allowed to realise a detailed study of piezoresistive pastes and to complete previous research in this field concerning the influence of firing parameters (quenching) and annealing studies. From a scientific point of view, this first step allowed to show that nanostructure, conduction mechanism and electrical properties are intimately linked. By choosing adequate and relevant compositions, structure and firing, we proposed a new way to unveil the conduction process that has not been yet elucidated. From a technical point of view, their stability could be enhanced with a higher GF or adapted TCR. However, they present a large range of applications because of their different Tf and their different TCRs. Thanks to this particularity, these pastes could be used on different substrates, and we could expect a larger technological impact by optimising our piezoresistive pastes by additives to better control their properties.