Monitoring the I-V parameters in mass production yields a distribution that cannot be understood in a simple manner. For example, if Voc varies greatly, it is not obvious whether this is mainly due to variations in the bulk lifetime or in the surface passivation or due to other sources. In this work, we develop a method where statistics is combined with numerical device modeling to obtain a physical interpretation of the observed variations. In the first part, we derive a multivariate statistical model to extract the main influences of fabrication fluctuations on the I-V parameters. This statistical model is based on cell parameters measured on a representative sample of solar cells from production. In the second part, we develop a computer-aided design (TCAD) device simulation model for multicrystalline Si solar cells. This TCAD model quantifies the I-V variations on a physically sound basis. However, the number of simulations is grossly reduced by feeding in solely the main influences obtained from the statistical model. In the third part, we verify this method by comparing the calculated distribution with production data. This model is used for optimization strategies for higher cell efficiency, smaller variations in cell parameters and improved yield in mass production. Furthermore, we will apply our methodology to advanced cell concepts. It will allow the early consideration of production fluctuation in device simulation of advanced cell concepts, and therefore a realistic assessment of such concepts.