Viscoelasticity and crack growth govern the long-term deformability of concrete and thus its service behaviour and its durability. For low load levels, viscoelasticity behaves quasilinearly and crack growth is inactive. On the other hand, for high load levels, cracks grow and interact with viscoelasticity. Numerous authors have long demonstrated the influence of microcracking on creep qualitatively. Moreover, rate effects on the fracture behaviour of concrete are clear for low and high loading rates. Nevertheless, the mechanisms involved in these effects are not yet clearly determined. Cohesive crack approaches extended to time dependent parameters, give some elements of explanation, although without providing a general tool capable of explaining all the phenomena. The dissipative nature of crack growth and viscoelasticity naturally encourages one to model their couplings by means of an energybased approach to fracture. This is the meaning of the continuum thermodynamics of dissipative multicracked granular bodies developed by Huet (1997). This theoretical formalism puts the emphasis on the effect of viscoelasticity on the driving (reactive) part of the propagation criteria (energy release rate). The aim of this experimental research was to investigate coupling effects between viscoelasticity and crack growth in concrete. With this aim in view, 4 different types of fracture test have been performed on the same material (concrete with a maximum aggregate size of 8 mm), with a constant specimen geometry (rectangular wedge splitting specimens 20 by 20 by 10 cm). The first type of test consisted of performing a series of successive relaxations, at various increasing load levels, before, on and after the peak force, following the envelope of failure. Special attention was devoted to the influence of the control parameter ("active": with respect to displacements measured on the specimen, or "passive": crosshead displacement) and loading history. The results showed a progressive deviation from the linear viscoelastic behaviour starting from a load level of approximately 50 % of the peak force, before the peak. This deviation was significantly higher with crosshead control ("passive"). It depended on the loading history. After the peak, in "active" displacement control, relative relaxations tended to be similar whatever the load level. The displacement parameters (other than control) measured on the specimen or crosshead during relaxation, showed an evolution dependent on the control parameter. This effect was significant at the beginning of the relaxation and tended to disappear later. Furthermore, acoustic emissions could be detected at the beginning and during some relaxations, for high load levels (nearby peak force and after). The second type of test consisted of one or more successive creep levels up to eventual failure under sustained force. In certain cases, during creep levels leading to fracture, alternating secondary and tertiary creep with multiple concavity changes correlated with measured acoustic emissions could be observed. The tertiary creep leading to fracture developed over several minutes. Unstable sudden crack propagation was only seen at the end of the tertiary creep levels, accompanied by a sudden and very fast increase of the number of acoustic emissions. Moreover, during tertiary creep, an increase of the crack length measured on the specimen's surface by means of a conductive crack graphite gauge could be observed, correlated with the evolution of the creep displacements. The third type of test was performed under constant displacement speed (displacement measured on the specimen in the axis of the splitting force) with a range of speeds from 5 x 10-4 to 5 x 10-1 mm/minute. The results show a clear influence of the displacement speed on the response in terms of force-displacement curve, in the vicinity of the peak. The maximum force decreases with the displacement speed. On the other hand, the displacement at peak force was not significantly influenced by the displacement speed. Surface crack length measurements by means of a conductive graphite gauge enabled the determination of the surface crack speed as a function of the imposed displacement. The obtained surface crack speed depended quasi linearly on the imposed displacement speed. The fourth type of test, complementary to the former ones, was especially for the study of the evolution of internal micromechanical parameters during crack growth. Two wedge splitting specimens were equipped internally, near to the potential crack path with optical strain gauges (Bragg gratings). The internal strain measurements, revealed two zones of very different behaviour, with similar tends for the two specimens. Optical strain gauges close to the fracture plane (less than 5 mm from gauge axis) measured high strains (around 1000 με) significantly greater than the ultimate tensile strain of normal concrete (0.1‰ or 100 με) and mostly irreversible. At the contrary, optical strain gauges farther from the fracture plane (over 10 to 15 mm) measured strains that were always smaller than 100 με , and mostly reversible. Finally, two types of computer simulation (homogeneous materials) were performed in order to complement the interpretation of experimental results. A linear viscoelastic calculation, for a constant crack pattern, was used to illustrate quantitatively the progressive deviation from linearity observed in the experimental successive relaxations. A non-linear calculation of the propagation of a discrete crack with softening behaviour, in a linear elastic material, accurately predicted some of the strains measured with optical strain gauges during crack propagation. The experimental results demonstrated many indications of crack growth activity during creep as well as relaxation levels. These measurements confirm the significant contribution of propagation and coalescence of microcracks to the non-linear viscoelastic response of concrete. Indications of microcrack growth during relaxation can be explained by unstable propagation phenomena triggered by the small size of the microcracks. These phenomena could be amplified by viscoelastic effects of the redistribution of internal stresses in the heterogeneous structure of concrete. All the work performed clearly shows the need to complement macroscopic measurements such as the reaction force of a specimen or external displacements, with the measurement of micromechanical parameters in order to distinguish the contributions of the different active phenomena and of their coupling.