Collagen is constituted of three polypeptides forming a triple helix maintained by numerous weak chemical interactions, such as hydrogen bonds. In connective tissue it further packs into very long cylinders of few nanometers of diameter, called fibrils. Once heated sufficiently, the molecules loose their original structure and take more random conformations (denaturation). The kinetics at which this reaction occurs is strongly dependent upon temperature, because of its co-operative nature. Although the reaction at the molecular level is likely complex and multiple, modelling the kinetics can be simplified considering a single reaction from the native to the denatured state. The rate limiting step of the reaction involves an activated state to which an activation enthalpy and entropy are attributed. The first part of this work is concerned with the characterisation of this activated state. Rates of denaturation are measured under isothermal conditions by monitoring the shrinkage of native bovine tendon. These measurements permit the extraction of the enthalpy and entropy of the molecular state limiting the kinetics. It is shown here that the model considering the activated state as the denatured state of the co-operative fraction of the molecule also permits a modelling of the effect of tension on the rate of reaction. Based upon the theory of rubber elasticity, we derive a quadratic dependence of the kinetics with linear tension applied to denaturing sample. This quadratic behaviour is verified experimentally. The effect of stress is also suspected to be responsible for the heterogeneous form of denatured tissue. By performing isothermal shrinkage measurements on rat tail tendons, the activation parameters of this tissue are also extracted. Additional cross-linking is realised by Glutaraldehyde fixation. It is found experimentally that additional cross-linking leads to a reduction of the entropy of the activated state. In the second part of this work, we are mainly interested in the kinetics of high temperature denaturation as induced by pulsed infrared lasers during surgical procedures. New tools are developed to monitor the reaction during and after CTH:YAG laser pulses on rat tail tendon. For this purpose, an imaging setup is built to generate images of the phase shift between two orthogonal polarisation directions of a probe beam passing through the sample. By careful analysis of the images obtained, the instantaneous local state of birefringence of the sample is derived. This very sensitive method is first used to analyse reversible structural changes of the tissue below the threshold of laser radiant exposure leading to denaturation. Surprisingly, we measured a totally reversible increase of birefringence with temperature. This mechanism is shown to be of different origin than the denaturation process and is suspected to be linked to a variation of the fibril diameter with temperature. The same optical method permits the observation of the state of denaturation of rat tail tendons that are heated by laser pulses of higher energy. Local sites of denaturation are detected by the irreversible modification of birefringence. The method allows one to determine the pulse threshold radiant exposure for rat tail tendon denaturation. Comparison is then made between the threshold radiant exposures determined experimentally with predictions made using the kinetic model whose parameters were obtained previously with isothermal measurements. This is our first successful attempt to validate the model for higher temperatures in a time scale of 10 ms, ruled by the heat diffusion. In order to characterise even higher rates of denaturation, the optical setup is slightly modified to permit a follow up of the reaction during the laser pulse. A streak imaging principle is added to the phase shift detection. The time scales probed by this technique are in the range of 50 μs. By applying higher power laser pulses on the samples, the method reveals a total denaturation in a fraction of a millisecond. The starting point of denaturation during a laser pulse is compared to the computations realised assuming a first order kinetic model. A good agreement between theory and experiments is observed. These experiments realised under high heating rates validate the use of the phenomenological first order kinetic model up to the 50 μs time scale.