One of the major challenges in drug development is the development of a drug-screening model that closely resembles the adult human heart. Currently many drugs are rejected in late stages of development and even withdrawn from the market due to possible cardiac tissue damaging side effects. The corresponding delay in drug development is mainly due to a lack in drug-screening methods, to determine these life-threatening side effects. Cultivation of human cardiac cells in vitro could provide such a drug-screening model. Human cardiac muscle cells can be derived from human embryonic stem cells. However, the cardiac muscle cells derived by culturing these human embryonic stem cells into cardiomyocytes (cardiac muscle cells) appear to be immature in relation to cardiac muscle cells taken directly from the adult human body. This immaturity results in a drug-screening model with limited predictability. The hypothesis of the Cytostretch project, a collaborative project between Leiden University Medical Centre and Philips Research, is that in vitro cultured cardiomyocytes subjected to in vivo mimicking mechanical stimuli (stretch), will show enhanced maturation. These mature cell cultures provide a good basis for future drug-screening models. A chip is developed containing a polydimethylsiloxane (PDMS) thin-film as a substrate for cardiomyocyte culturing. A pressure difference is applied to the membrane to obtain a strained membrane, which with anchored cardiomyocytes will result in cardiomyocyte stretch. In order to stretch cardiac myocytes in vitro, the development of an in vivo mimicking loading protocol is essential. The main goal for the presented study, therefore, is the development of a proper loading protocol for in vitro stretching of cardiomyocytes. To reach this goal two study objectives have been defined; investigation of in vivo cardiac muscle strain during a normal cardiac cycle and the determination of the PDMS thin-film behavior. For the determination of the in vivo cardiac muscle strain, the left ventricle was modelled mathematically. Assumed was a homogeneous stress distribution along the left ventricular wall, corresponding to a fiber direction course along the wall, consistent with anatomical findings. The relation between left ventricular fiber stress and left ventricle pressure showed to depend mainly on the ratio of cavity volume over wall volume. The left ventricle mechanics can be approximated by: Where ?f is the left ventricular fiber stress, Plv the left ventricular pressure. Vlv is the left ventricular cavity volume, Vw the left ventricular wall volume and ??f the natural fiber strain. The outcome of the mathematical left ventricle model led to the conclusion that the absolute left ventricular fiber strain between end systole (reference volume) and end diastole equals an approximate 14.7% for a healthy adult human heart during normal cardiac cycle. The PDMS thin-film behavior is modelled analytically with use of classical thin plate mechanics, considering large deformations. The analytical derived outcomes were subsequently compared with numerical and experimental results. The thin-film mechanics appeared to depend mainly on strain due to the extension of the membrane, and only little on strain related to bending. Moreover, the bending strain could be neglected when a pressure of 3kPa or more was applied to the membrane. It was concluded that an analytical model, simply supported around the edges, assuming a linear elastic homogeneous isotropic material, describes the membrane behavior properly. An analytical model with clamped edges was not able to deal with the small radius curvature at the edges due to the great flexibility of the membrane. The displacement field corresponding with the supported boundary condition was consistent with both the numerical approximation and the experimental data. The results of the analytical model showed that the order of the in plane displacement function u has major influence on the strain outcomes of the model. An in plane displacement function with two terms showe d a great strain variation across the membrane, whereas an in plane displacement function containing 5 terms showed a relatively homogeneous strain distribution, consistent with the numerical approximation. From the results of the analytical model the conclusion was drawn that the pressure-strain behavior of the two membrane configurations (circular and dogbone) differs. In order to obtain a maximum membrane strain of 14.7%, the applied pressure for the dogbone membrane should be 3.725kPa, whereas for the circular membrane the applied pressure should be 5.375kPa. Moreover, on the circular membrane, the transverse strain differs from the radial strain. The radial strain showed to be homogeneously distributed over the entire membrane, resulting in longitudinal equally stretched cardiomyocytes. The transverse strain however decreased from the centre outwards. In the centre section this will result in cells equally stretched longitudinal as well as transverse, however in the edge sections the cells will receive considerably lower stretch in the transverse direction. Some preliminary testing has been performed in stretching beating areas (clogs of cardiomyocytes) in vitro while applying the amount of pressure to the membranes as described above. The first set of experiments showed a maintained cell anchorage to the moving substrate for a long period of time (>120 hours). The second set of experiments showed detachment of the cells at increased pressures up to approximately 7kPa (for both configurations). From this we were able to conclude that a moving substrate has no detrimental impact on cardiomyocyte anchorage. Furthermore, from the detachment of cells at higher pressures than derived we are able to conclude that the pressure protocol will be in a correct range. The fact that the moving substrate has no detrimental impact on the cardiomyocytes indicates the great opportunities for the Cytostretch project.